Glossary of Audio and Video Archiving Terms

This page collects the technical terms used across the audio and video archiving guides on this site. It’s alphabetical, with each entry deep-linkable — you can refer to a specific term from anywhere with a URL like /glossary-of-terms/#tbc. Where two terms mean the same thing, you’ll find a cross-reference rather than two copies of the same definition.

0-9 · A · B · C · D · E · F · G · H · I · J · K · L · M · N · O · P · Q · R · S · T · U · V · W · X · Y

0-9

3-2-1 backup — A standard data-redundancy discipline: three copies of every file, on at least two different storage media, with at least one copy stored off-site. The minimum sensible posture for an archive that is meant to survive longer than the original storage hardware.

3-head cassette deck — A cassette deck with separate dedicated heads for erase, record, and playback. The third (playback) head is widely considered to improve playback fidelity because it can be optimised for one job rather than compromised between record and playback duties.

3.5mm jack (1/8″) — The smaller of the two common headphone connectors. Found on phones, MP3 players, camcorders, and most consumer audio gear. Unbalanced.

4fsc — A sampling rate of four times the colour subcarrier frequency. For NTSC, 4fsc is 14.318 MHz; for PAL, 17.734 MHz. Used by digital composite video formats (D-2, D-3) and by vhs-decode’s intermediate .tbc files, which sample at 4fsc to represent the composite signal faithfully.

6.35mm jack (1/4″) — The bigger of the two common headphone connectors. Found on broadcast, radio, and studio gear; almost universal on studio headphones. Can be wired as unbalanced (TS) or balanced (TRS).

A

ADC (Analog-to-Digital Converter) — A circuit that converts a continuous analog signal into a series of digital samples. The chip at the heart of any analog capture card or audio interface.

A-D strips — Acid-detection strips used to monitor film for vinegar syndrome. The dyed strip changes colour in the presence of acetic acid, giving an early warning that acetate film is decaying so it can be prioritised for cold storage or scanning.

AGC (Automatic Gain Control) — A circuit that automatically adjusts signal levels to a target range. In VHS playback, AGC can be triggered (intentionally or otherwise) by Macrovision-style pulses, producing rolling or darkening pictures.

AIFF (Audio Interchange File Format) — Apple’s uncompressed audio file format from 1988. Functionally identical to WAV for archival purposes — both store raw PCM audio with no compression, both are lossless.

Anti-static brush — A record-cleaning brush, usually carbon-fibre, that lifts surface dust while bleeding off the static charge that attracts more of it. Used dry, before each play.

Archival master — The lossless preservation copy of a capture, kept untouched so future tools and workflows can be applied to it. Distinct from a viewing copy, which is the playable derivative.

Audio channel — A single, mono stream of audio. Two channels make stereo; six make 5.1 surround.

AV1 — A modern royalty-free video codec, more efficient than H.265, slowly gaining hardware support. Not yet typical for archival use.

AVC (Advanced Video Coding) — Another name for H.264.

Azimuth — The angle of a tape head relative to the tape’s direction of travel. Misalignment between record-time azimuth and playback-time azimuth causes loss of high frequencies in audio, and chroma in video.

B

Baking — A controlled low-temperature treatment (typically several hours in a food-dehydrator-style oven) that temporarily reverses sticky-shed syndrome in magnetic tape, letting a degraded tape be played and captured once before it relapses. See sticky-shed syndrome.

Balanced audio — A wiring scheme that reduces noise on long audio cables. Uses three conductors per signal (signal, inverted signal, ground); the receiving equipment subtracts the inverted from the non-inverted, cancelling any noise picked up equally on both. Found on XLR and 3-pole TRS connectors.

Baseband — A signal at its native frequency, near zero hertz, not modulated onto a higher-frequency carrier. Baseband audio is the waveform at a deck’s RCA outputs after the deck has decoded whatever was on the tape — whether the source on the tape was the linear analog audio track along the edge of the cassette, or the FM-modulated HiFi carriers that the deck demodulated internally. Composite video at the yellow RCA is also baseband. Contrast with a modulated signal, where the original has been encoded onto a high-frequency carrier — VHS HiFi RF at around 1.4 megahertz on the helical drum, the FM-modulated luma carrier in the video RF, or a radio broadcast. The term refers to frequency, not voltage; “baseband” doesn’t mean zero volts. See: Capture hardware in 2026.

Bathtub curve — The characteristic failure-rate pattern of hard drives over time: higher early (infant-mortality) failures, a long low plateau through the middle years, then a rising rate as the drive ages. Named for the shape of the curve.

Betamax — Sony’s consumer videocassette format — the contemporary rival that lost the format war to VHS despite comparable quality. Colour-under, standard definition, and increasingly dependent on scarce decks. Distinct from the professional Betacam family.

Binder hydrolysis — The chemical breakdown of the binder, the glue that holds the magnetic particles onto a tape’s backing, as it slowly absorbs moisture from the air. It is the main thing limiting how long magnetic tape survives, and its advanced, gummy stage is sticky-shed syndrome. Warm, damp storage speeds it up. See sticky-shed syndrome, baking.

Bit depth — The number of bits used to represent each sample. 16-bit gives 65,536 possible values per channel (CD quality); 24-bit gives ~16.7 million; 32-bit float covers an even wider range with finer detail. Higher bit depth gives more headroom for grading or restoration.

Bit rate — The number of bits used per second. For audio, calculated by multiplying bit depth × sample rate × channels. For video, depends on codec and content complexity.

Bit rot — Slow corruption of stored data over time, due to magnetic decay, flash-cell leakage, optical-layer breakdown, or simple medium failure. The reason archives need integrity checks and redundant copies.

Breathing (pumping) — An audible rise and fall in background noise level as a companding noise-reduction system’s gain reacts to the music, most noticeable on quiet passages right after a loud one. A sign of a mismatched or miscalibrated encode/decode pair. See companding.

Brickwall compression — Aggressive peak limiting that flattens a recording’s loudest peaks against a hard ceiling to make it as loud as possible. A hallmark of the loudness war, it reduces dynamic range.

bwdif (Bob Weaver Deinterlacing Filter) — An open-source FFmpeg deinterlacing filter. Produces decent results on most material at low computational cost. QTGMC produces noticeably better output but is much slower. Worth knowing about because Topaz Video AI uses bwdif internally as its deinterlacing pre-processor — the AI branding in Topaz is on the upscale or denoise stage that runs after deinterlacing, not on the deinterlacer itself.

C

Capstan — The motor-driven shaft inside a tape deck that pulls the tape past the heads at a controlled, constant speed. Capstan health is essential for stable playback; a worn capstan or pinch roller will produce speed instability and wow.

Capture card — A device (PCIe, USB, FireWire, or Thunderbolt) that takes an analog or digital video signal and writes it to disk as a file. Quality varies enormously between brands and tiers.

Carbon-fibre brush — A vinyl-cleaning brush whose fine conductive bristles reach into the groove to lift dust while dissipating static. The standard light, dry clean to use before every play.

Cassette deck — A device that plays (and often records) audio cassette tapes. Comes in many forms — portable Walkman-style, rack-mount studio decks, separates for home Hi-Fi systems. The deck handles the mechanical transport of the tape past the playback heads.

CCD (Charge-Coupled Device) — A type of image sensor common in high-quality scanners and older digital cameras. Light hits the sensor and gets converted to an electrical charge that’s read out row by row. CCD sensors generally have better quality — wider dynamic range, less noise — than the cheaper CIS sensors used in printer-scanners.

CD bronzing — A form of disc rot in which the reflective layer of certain early pressed CDs corrodes and discolours to a bronze tint, eventually making the disc unreadable. An oxidation problem, not dye fade.

CEA-608 / EIA-608 — The standard for line-21 closed captions on analogue NTSC video, the same data format used from the original 1980s decoder chips through to line-21 captions on DVD. See closed captions, VBI.

Cellulose acetate — The plastic base used for most film and some early tape from roughly the 1950s to the 1980s, replacing flammable nitrate. It is prone to vinegar syndrome as it ages. The later polyester base is far more stable. See vinegar syndrome, polyester base, nitrate film.

Channel separation — How completely the left and right stereo channels are kept apart. On vinyl both channels share one groove, so separation is limited to roughly 20 to 30 dB; in digital the channels stay discrete and separation exceeds 90 dB.

Chroma — The colour information in a video signal, as distinct from luminance (brightness). Stored separately in nearly all video formats because the eye is less sensitive to chroma detail.

Chroma subsampling — A method of reducing the amount of data needed to represent chroma, by storing chroma at lower resolution than luminance. Common ratios are 4:4:4 (full chroma), 4:2:2 (half horizontal chroma), 4:2:0 (half horizontal and vertical chroma) and 4:1:1 (quarter horizontal chroma). See: Understanding 4:2:2 chroma subsampling.

Cinching — Slippage between the layers of a wound tape pack, which can crease or fold the tape. Encouraged by storing cassettes flat rather than upright. Also called pack slip.

Cineform — A line of professional intermediate codecs from GoPro (originally CineForm Inc.). Visually lossless at the higher quality grades but not mathematically lossless — discards data, just not enough for the difference to be visible at one generation. Used widely as a Windows-side editing intermediate in colour-graded film and video work. Not an appropriate archive format for the same reason ProRes isn’t: the decoded pixels are close to but not bit-identical to the input, and the format is proprietary. Treat as a viewing or delivery codec, not a master.

CIS (Contact Image Sensor) — A simpler image sensor type used in multifunction printers, sheet-feed scanners and other compact devices. The sensor sits right against the glass — hence “contact” — which is cheaper and faster than CCD but produces flatter images with less dynamic range and a very shallow depth of field. Anything thicker than a flat photo loses detail quickly.

Client-side encryption — Encrypting your data yourself, on your own machine, before it is uploaded anywhere, with a tool such as Restic, Borg or rclone. The only arrangement where you provably hold the key, because you generated it, and it turns any plain storage into a private one whoever hosts it. See Keeping your family archive out of the wrong hands.

Clockgen Mod — A small daughter-board modification for CX-card based RF capture rigs. It ties the audio and video sampling clocks together, fixing a problem where independent clock crystals drift slightly against each other over a long capture and produce audio-video sync issues you only notice afterwards. See: Capture hardware in 2026.

Cold storage — Object storage priced for data you will almost never read back, such as Amazon’s Glacier Deep Archive and the Archive tiers from Azure and Google, at around a dollar per terabyte a month. The catch is retrieval, which takes hours, costs a fee and runs through developer tooling, making it a poor fit for a family archive. See Choosing a cloud provider.

Colour burst — A short reference pulse sent just before each active video line, giving the receiver’s colour decoder a fixed phase to lock onto. Without it a composite or S-Video picture loses colour sync and the hue drifts or disappears. See colour-under.

Colour dye fading — The gradual loss of colour in film and photographs as the chemical dyes that form the image break down, driven mostly by heat and humidity rather than light. Black-and-white images, made of metallic silver rather than dye, are far more stable. Cool, dry storage slows it. See cold storage.

Colour subcarrier — The higher-frequency signal (3.58MHz in NTSC, 4.43MHz in PAL) that carries colour information layered on top of the black-and-white picture in a composite or colour-under signal. Separating it cleanly from the luminance is the whole job of a comb filter. See comb filter, colour-under.

Colour-under — The method of recording colour information on tape used by VHS, Betamax, U-matic and similar consumer/prosumer formats. The colour signal is shifted down to a low frequency (around 629 kHz for NTSC, 627 kHz for PAL) before being recorded onto the tape, then shifted back up to the standard subcarrier frequency on playback. This is why VHS chroma has so little bandwidth, and why colour quality is the format’s biggest weakness. See: How vhs-decode actually works.

Closed captions — Text overlays carrying dialogue or descriptive audio for accessibility. On NTSC, transmitted on Line 21 of the VBI; on PAL broadcasts, typically through teletext. Discarded by most consumer capture workflows but preserved by RF capture.

Comb filter — A filter that separates the luminance and chroma components of a composite signal, used inside many VCRs and capture chains. Quality varies enormously; a bad comb filter produces chroma noise and rainbow artefacts on saturated edges.

Companding — Compressing a signal’s dynamic range before recording, then expanding it back out on playback, so tape hiss added during recording gets pushed back down by the expansion. The mechanism behind Dolby B/C/S and dbx noise reduction; decoding with the wrong setting (or none) is what makes a tape sound muffled or overly hissy. See Dolby NR.

Component video — A video signal split into three separate channels (typically Y, Pb, Pr) carried on three cables. Higher quality than composite or S-Video. Not used on consumer VHS but appears on broadcast and professional gear.

Composite video — A video signal where luminance and chroma are combined into a single waveform on one wire — the yellow RCA connector on most consumer gear. The simplest and lowest-fidelity of the common analog video formats.

Compression — A process for making files smaller. Lossy compression discards information considered unlikely to be noticed; lossless compression discards nothing and decodes back to the bit-identical original.

Container format — A file wrapper that can hold one or more video, audio, subtitle, and metadata streams. Common containers are MKV, MP4, MOV, and AVI. Distinct from the codec, which is how the data inside is compressed.

Control track — A fixed pulse track running along the bottom edge of the tape, laid down at recording time. A VCR’s capstan servo uses it to keep playback speed locked to the original recording, and a worn or misaligned control track is one of the causes of tracking lines. See tracking.

CRF (Constant Rate Factor) — An encoding setting in modern video codecs (x264, x265, AV1) that targets a perceptual quality level rather than a fixed file size. Lower numbers mean higher quality and bigger files; higher numbers mean smaller files and more visible artefacts. CRF 18 is generally considered visually lossless for SD analog material; CRF 20-23 is typical for delivery copies.

Cross-colour — Interference where fine picture detail (a herringbone jacket, venetian blinds) gets misread as colour information, producing a rainbow shimmer. The composite-video sibling of dot crawl, and one more reason a comb filter or S-Video connection gives a cleaner picture. See comb filter, dot crawl.

CRT (Cathode Ray Tube) — The original electronic display technology, using an electron beam to excite phosphors on a glass screen. Standard for video and computer displays before LCDs; still used for some calibration work because of how cleanly they handle motion.

cvbs-decode — A software-decode project for composite (CVBS) video sources captured directly at the source. Sister project to vhs-decode and ld-decode. Still maturing as of 2026, but covers the use case where the source is already composite and capturing it with high sample rates is more useful than running it through a conventional capture card. See: How vhs-decode actually works.

CXADC — A Linux driver and capture mode for certain Conexant CX2388x-based capture cards, allowing them to record raw uncompressed sample streams. Used in vhs-decode workflows for capturing baseband or RF signals. See: Capture hardware in 2026.

D

DAC (Digital-to-Analog Converter) — A circuit that converts a digital signal into an analog one. Found in CD players, audio interfaces, and the output stages of video equipment.

Dark fading — The gradual fading of colour-film and colour-photo dyes that happens even in total darkness, driven mainly by heat and humidity rather than light. The cyan layer usually goes first, leaving a pink or magenta cast.

Data migration — Copying digital files onto fresh storage every few years, before the old drive, disc or format fails or falls out of use. Because no storage medium lasts forever, migration, not any single durable disc, is what keeps digital files alive. See bit-rot, fixity, 3-2-1 backup.

Data retention — How long a storage medium holds its data without loss. For an unpowered SSD this is finite: the trapped charge in each cell leaks away over months to years, faster when warm.

DAW (Digital Audio Workstation) — Software for recording, editing, and mixing audio. Examples include Adobe Audition, DaVinci Resolve’s audio side, OcenAudio, Audacity, Pro Tools, and Reaper.

DCT (Discrete Cosine Transform) — The mathematical transform at the heart of most lossy video codecs (JPEG, MPEG-2, H.264, ProRes). The codec converts each block of pixels into frequency components and quantises the less-visible high-frequency components more aggressively, which is how it discards data.

ddd-capture-toolkit — Open-source batch workflow software for the full vhs-decode pipeline: capture, decode, compress, export, audio align, mux, validate, archive. Grew out of fixing the DomesdayDuplicator’s audio-sync problem and now coordinates the entire chain — queuing, prioritising and tracking the multi-step work across many tapes — for any setup that can produce a raw capture file, with or without DdD hardware. Runs in a terminal on a headless Linux box, so the typical use is a dedicated machine working through a collection overnight. Cryptographically validates that the compressed archive copies match the original captures. See: Capture hardware in 2026.

Deacetylation — The chemical breakdown of cellulose-acetate film base, in which the acetate hydrolyses and releases acetic acid. The underlying reaction of vinegar syndrome.

Deionised water — Water with its mineral ions removed, used like distilled water for cleaning records because it leaves no mineral residue in the groove. Tap water does the opposite.

Delamination — The separation of the bonded layers of a disc or tape, such as the lacquer and reflective layers of an optical disc coming apart. A common end stage of disc rot and of tape decay.

Diffusion model — A category of generative AI that synthesises content by progressively removing noise from a random starting point, guided by the input. Diffusion models can produce strikingly realistic images and video and are the technology behind several recent video tools including Topaz Starlight and ByteDance’s SeedVR2. For analog video restoration they share the structural property of all super-resolution work — they produce plausible detail rather than recovered detail.

Deck — Shorthand for a video or audio tape player/recorder. “VCR” is the consumer term; “deck” is the serious-user equivalent.

Demagnetiser — A tool that removes unwanted residual magnetism from tape heads and ferrous components. Stray magnetism causes noise and gradual erasure of magnetic media.

Digital audio interface — A device that handles the conversion between analog audio (from a tape deck, microphone or instrument) and digital audio (in your computer). It contains an ADC for input and a DAC for output, and connects to the computer over USB, Thunderbolt or PCIe. The same kind of device you’d use for recording a podcast or producing music.

Digital noise reduction (DNR) — Circuitry in better VCRs and capture devices that reduces video noise in the signal. Useful for clean capture, though heavy application can soften fine detail.

DIN — A family of circular multi-pin connectors once common on European audio gear, including some reel-to-reel and cassette decks. A 5-pin DIN can carry stereo in and out on a single plug; connecting an older deck to a modern interface usually means a DIN-to-RCA lead.

Disc rot — The degradation of a CD, DVD or Blu-ray as its recording dye breaks down or its reflective layer corrodes, until the disc can no longer be read. It is often invisible until the disc fails, which is why recordable discs are not a set-and-forget archive. See bit-rot, M-Disc.

Distilled water — Water purified by boiling and condensing, leaving minerals behind. Preferred for wet record cleaning because it dries without leaving deposits in the groove.

Dolby Digital — A multi-channel audio compression standard used on DVDs, broadcast, and some streaming. Not the same thing as Dolby NR (noise reduction).

Dolby NR (Noise Reduction) — A family of analog noise reduction systems used on cassette and reel-to-reel tape. Dolby B, C, S, and HX-Pro each reduce tape hiss in slightly different ways; they must be decoded on playback with the same setting they were recorded at.

Domesday Duplicator (DdD) — A hardware device built by the vhs-decode community that captures the raw RF signal from a VCR head, allowing software-based decoding instead of relying on the VCR’s own playback electronics. See: Capture hardware in 2026.

Dot crawl — A crawling checkerboard pattern along sharp vertical colour edges in a composite-video picture, caused by the luminance and colour subcarrier signals interfering with each other. Reduced by a comb filter, and avoided entirely on S-Video or component connections, which keep luma and chroma on separate wires. See comb filter, cross-colour.

DPI (dots per inch) — A resolution unit, originally from printing where one “dot” was one ink dot per inch of paper. For scanners it refers to how many image samples are taken per inch of the original. Higher DPI means more detail captured — but only up to the limit of the sensor’s actual physical resolution (see optical resolution).

DPX (Digital Picture Exchange) — A 10-bit log image format used in film post-production. Each frame of a scanned film negative is typically stored as a DPX file, with the colour values encoded logarithmically so that shadow and highlight detail is preserved without crushing. Common in cinema-grade workflows.

Dropout — A momentary loss of signal during tape playback, caused by oxide loss, debris in the head path, or a brief loss of head contact. On VHS the FM carrier briefly disappears; the signal defaults to peak white, producing a white horizontal flash.

DSLR (Digital Single-Lens Reflex) — A camera body design where you look through the same lens that takes the picture, via a mirror that flips up out of the way during the exposure. The dominant serious-photography category from roughly 2000 to 2015, before mirrorless cameras took over. DSLRs and mirrorless cameras (which work the same way for archival purposes) are increasingly used as makeshift film scanners — the camera shoots a backlit negative through a macro lens, and software inverts the image.

DV (Digital Video) — A consumer digital video format from the late 1990s, intra-frame compressed at 25 Mbit/s. Transferred to computer over FireWire as a bit-identical copy of what’s on the tape; the encoding was done at record time.

DVCAM — Sony’s prosumer variant of DV, using a slightly faster tape transport. Same DV25 codec; plays back in any DV-compatible deck.

dvgrab — A Linux command-line tool for capturing DV (digital video) from MiniDV camcorders and similar devices over FireWire. The Linux equivalent of WinDV; often more reliable than the Windows tools on modern systems.

DVRescue — Open-source software (from MIPoPS and MediaArea) for migrating DV tapes to preservation files. Its standout feature is automatic error recovery: over FireWire it controls the deck and, when it detects dropped or errored frames, rewinds and re-captures them — which rescues marginal or lightly-damaged tapes. Supports DV, DVCAM, DVCPRO, MiniDV and Digital8 (but not HDV). See dvgrab, vrecord.

DWPD (drive writes per day) — An SSD endurance rating: how many times the drive’s full capacity can be written each day across its warranty period. A complementary measure to TBW.

Dynamic range — The difference between the brightest detail and the darkest detail a system can capture without losing information. A camera or scanner with wide dynamic range can record both bright highlights and deep shadows in the same image; a narrow-dynamic-range device crushes one or both into uniform black or white. Measured in stops (for cameras) or decibels (for audio).

E

Eastmancolor — Kodak’s colour film process introduced in 1950, whose early dyes were unstable. Many 1950s prints survive only as faded magenta, making Eastmancolor a byword for colour-dye fading.

Ektachrome — Kodak’s consumer reversal film (slide film) from the 1960s onward. Unlike Kodachrome, its dye layers are not stable over decades; magenta-layer collapse causes cyan or pink shifts on old Ektachrome reels. Transfer Ektachrome sooner rather than later.

Dynamic DNS (DDNS) — A name that keeps pointing to your home internet connection even as your provider changes its address, so a machine elsewhere can find home reliably. One of the ways, with a fixed IP or IPv6, to reach your own NAS for off-site replication without leaning on a maker’s cloud service. See Self-hosting your off-site backup.

EDID (Extended Display Identification Data) — The handshake protocol used over HDMI and DisplayPort, where a display tells the connected device (computer, capture box) its supported resolutions and refresh rates. Sometimes an HDMI capture device needs special “EDID emulation” to convince a player that it’s a TV and to keep outputting video.

Egress — What a cloud provider charges for data leaving its storage — i.e. for downloading your files back out. It’s easy to miss because it doesn’t show on the storage price, but on a large archive the egress bill for a full restore can dwarf a year’s storage fees.

Encoding format — The codec used to compress video or audio inside a container. Often confused with container format; the same container (e.g. MP4) can hold many different codecs.

Encrypted at rest — Data stored in scrambled form on the provider’s disks, so a stolen drive yields gibberish. Standard, and not the point that matters most. The real question is who holds the key: provider-held keys mean the company, a court order or a breach can still read it. See Keeping your family archive out of the wrong hands.

Encrypted in transit — Data scrambled while it travels across the internet, so it cannot be read in flight. Standard on any reputable service. It protects the journey, not the destination, and says nothing about who can read the data once it has arrived. See Keeping your family archive out of the wrong hands.

End-to-end encryption (E2EE) — Encryption applied on your device before data leaves it, with a key the provider never sees, so only the two ends can read the content. The basis of any real zero-knowledge claim, and worth the same suspicion, since the encrypting software is usually the provider’s and closed. See Keeping your family archive out of the wrong hands.

Equilibration — Letting a cold tape, film or disc warm up gradually to room temperature, still sealed, before use, so that moisture does not condense on it. Also called acclimatization.

EXR (OpenEXR) — A high-dynamic-range image format developed at Industrial Light & Magic in 1999. Supports 16-bit and 32-bit floating-point per channel — much higher precision than standard 8-bit or 16-bit integer formats. Standard for digital cinema and high-end visual effects work.

F

FFV1 — A mathematically lossless video codec, open and patent-free as a standard FFmpeg codec, used as the standard archival codec by many national archives. Stores built-in integrity checks (checksums) for each segment of every frame, so if part of the file corrupts the codec can tell you exactly which segment failed.

Field — Half a frame of interlaced video — either the odd-numbered or even-numbered lines, taken at one moment in time. NTSC has 59.94 fields per second; PAL has 50. Two fields interleave to make one displayed frame.

FFmpeg — A free, open-source command-line tool that handles almost every audio and video codec, container format, filter and stream operation imaginable. Powers most other audio/video software underneath. If you ever do anything beyond a single click in a media app, you’ll eventually hit a problem that FFmpeg can solve.

FireWire (IEEE-1394, i.Link) — A high-speed serial connection used for DV, DVCAM, HDV, and Digital8 capture. Mostly absent from modern PCs, which has become the main practical bottleneck in MiniDV digitisation.

First pressing — A copy from the earliest manufacturing run of a record, made closest to the original master. Generally the most sought-after and valuable version of a title, identified by matrix etchings and label details.

Fixity — Checking that a digital file has not silently changed or corrupted over time, usually by comparing a checksum, a short fingerprint of the file, against one taken earlier. Regular fixity checks are how archives catch bit-rot before it spreads through every copy. See bit-rot, data migration.

FLAC (Free Lossless Audio Codec) — A lossless, open audio codec. Technically both a codec and a container, but normally referred to as a codec. The de facto archival format for digital audio and at level 6/8 the most effective way to store FM RF captures.

Flash memory — The solid-state storage inside SSDs, USB sticks and memory cards. It holds data as trapped electrical charge, which slowly leaks away when the device is left unpowered, so flash is a poor choice for long-term, unplugged storage. See SSD, hard disk drive.

Flatbed scanner — A scanner where the original sits face-down on a glass surface, scanned by a sensor head that moves underneath. Standard form factor for paper photos, books, and (with a transparency adapter) film negatives and slides. Contrast with sheet-feed scanners, which pull the original through past a stationary sensor.

FM RF (Frequency Modulated Radio Frequency) — A way of encoding a signal by varying the frequency of a carrier wave. Used for HiFi audio on VHS (also called AFM) and for video itself (VHS/Betamax/8mm is colour-under, LaserDisc & Type-C is composite).

FM RF Archival – The act or actions of preserving a soruce video, audio or data signal on an analouge meidum in standard raw PCM samples or FLAC compressed samples – term coined by Harry Munday in the 2020s (vhs-decode project)

Format obsolescence — When the machine, software or file format needed to read a recording is no longer made or supported, so the content becomes unplayable even though the carrier itself is fine. For tape this is often a bigger threat than decay, because you cannot manufacture a new VCR. See VCR.

Four-head VCR — A VCR with four video heads rather than two, giving cleaner playback and steadier still and slow-motion. A feature worth looking for in a deck used to digitise tapes.

Frame — A complete picture in a video signal. In interlaced video, a frame is made up of two fields shown 1/50th or 1/59.94th of a second apart.

Frame rate — How many frames per second a video runs at. NTSC video is 29.97 fps (or 59.94 interlaced fields); PAL is 25 fps (50 fields). Film is typically 24 fps.

Frame synchroniser — A device that locks an incoming video signal to a stable internal reference, stabilising its timing enough for downstream equipment to track it cleanly. Often confused with a TBC — many devices marketed as “TBCs” are actually frame synchronisers with a proc-amp attached. See: TBC vs Frame Sync.

G

Gamma — The curve that maps stored signal values to actual screen brightness. Designed to roughly match how the human eye perceives brightness — the eye is more sensitive to changes in shadows than in highlights, so gamma concentrates more detail down at the dark end. Different gamma standards (Rec. 601, Rec. 709, sRGB, BT.1886) describe slightly different curves; mismatched gamma between encoding and display causes brightness and saturation shifts.

Gamut — The range of colours a system can represent. Different colour spaces (Rec. 601, Rec. 709, P3, Rec. 2020) cover different gamuts. VHS has a relatively narrow native gamut.

Gate weave — Slight horizontal or vertical wobble of film as it passes through a projector or scanner gate, visible as the image moving subtly within the frame. Removable in post with stabilisation tools.

Generative AI — Software that produces new content (text, images, audio, video) by sampling from patterns learned during training, rather than retrieving or transforming existing content. In image and video work, generative AI underlies most “AI restoration” and “AI upscaling” tools, including Topaz Video AI’s newer models and the wider landscape of diffusion-based tools. Generative output is plausible but not evidentially faithful to a specific source; for archival purposes that distinction matters.

Generational loss — Cumulative quality loss when content passes through multiple encoding or copying steps. Especially visible with lossy codecs, repeated chroma-subsampling cycles, or repeated RGB ↔ YCbCr conversions.

Goldmine grading — The eight-step condition scale, Mint down to Poor, that the record trade and Discogs use to grade vinyl. Condition multiplies value: a Very Good copy is worth roughly a quarter of a Near Mint one.

GOP (Group of Pictures) — In inter-frame codecs (MPEG-2, H.264, H.265, AV1), a sequence of frames that begins with a keyframe and continues with predicted frames depending on the keyframe. Long-GOP codecs are more efficient but harder to edit than intra-frame codecs.

H

H.264 / MPEG-4 AVC — A widely used video codec with broad hardware support. Used for streaming, broadcast, web video, and Blu-ray. Less efficient than H.265 but more universally compatible.

H.265 / HEVC — A more efficient successor to H.264, achieving similar visual quality at roughly half the bitrate. Broad hardware support today, though the rollout was historically slow due to patent licensing complications.

H.266 / VVC — The next-generation codec after H.265, offering a further 30-50% bitrate reduction at the same quality. Very limited hardware support as of 2026.

HDV (High-Definition Video) — A high-definition format recorded on MiniDV-shaped tapes using MPEG-2 Long-GOP compression at 25 Mbit/s. Physically identical cassettes to MiniDV, but requires HDV-aware capture software.

Head alignment — The physical positioning and angle of a tape head. Misalignment causes loss of high frequencies in audio and chroma in video. Professional tape transfer often involves checking and adjusting head alignment for each tape.

Head amplifier — A small PCB inside a VCR sitting near the rotating head drum. It’s the first stage of electronics the heads’ output passes through, taking the very small RF signal coming off the spinning heads and amplifying it for the rest of the deck. The test points on the head amplifier expose the raw FM signal coming off the tape, which is what a vhs-decode capture rig taps into. See: How vhs-decode actually works.

Head switching — The point in a VHS frame where the helically scanned drum hands over from one head to the other. Appears as a band of noise at the very bottom of each frame. CRTs hid it in overscan; flat panels show it.

Hard disk drive (HDD) — A storage device that records data magnetically on spinning platters. Dependable as a working copy, but with a service life of only a few years and a real failure rate, so it belongs in a backup plan as one copy, never as an archive on its own. See SSD, 3-2-1 backup, RAID.

High-bias — A cassette-tape formulation (chrome or cobalt, Type II) that needs more recording bias than standard ferric tape, giving lower noise and better treble. Sealed high-bias blanks have a small collector market.

High definition audio — A consumer term for audio above CD quality. CD is 16-bit, 44.1 kHz; HD audio typically starts at 24-bit, 96 kHz.

Helical scan — The recording method used by VHS and most videotape, where the heads sit on a spinning drum tilted to the tape and lay the picture down in long diagonal stripes. It packs a lot of signal onto slow-moving tape, at the cost of fussy tracking and tension. See tracking, head-switching.

Hi-Fi audio (AFM) — The high-quality audio track on VHS and Betamax tapes, encoded as a frequency-modulated FM signal on the same head as the video. Distinct from — and significantly better than — the linear audio track that also exists on the tape. Most archivists want to capture the Hi-Fi track because the linear track is usually noisy and limited.

hifi-decode — A software-decode project for the FM-modulated HiFi stereo audio tracks on VHS, Betamax and 8mm tapes. Reads the captured RF — either from the same capture as the video, or from a separate audio-head tap — and produces a digital audio file. Sister project to vhs-decode. See: How vhs-decode actually works.

HuffYUV — An older mathematically lossless video codec, 8-bit. Largely superseded by FFV1, which is more efficient and supports higher bit depths.

Hygroscopic — Readily absorbing moisture from the air. The polyurethane binder of magnetic tape is hygroscopic, which is why humidity drives binder hydrolysis and sticky-shed.

I

IEEE-1394 — See FireWire.

i.LINK — Sony’s brand name for FireWire (IEEE-1394). See FireWire.

Inter-frame compression — A class of video compression that encodes some frames as differences from neighbouring frames. More efficient than intra-frame but harder to edit and less resilient to packet loss. Examples: MPEG-2, H.264, H.265, AV1.

Interlacing — A video scanning method that draws odd-numbered lines first, then even-numbered lines, in two separate fields per frame. Used by NTSC, PAL, and all standard-definition broadcast formats. Modern displays are progressive and have to deinterlace interlaced material for display.

Intra-frame compression — A class of video compression where each frame is encoded independently. Less efficient than inter-frame but more editable and more resilient to packet loss. Examples: ProRes, DV, FFV1, JPEG sequences.

I-frame, P-frame, B-frame — In modern video codecs (MPEG-2, H.264, H.265, AV1), three kinds of frame:
I-frame (Intra) — a complete, self-contained frame, like a JPEG.
P-frame (Predicted) — only stores the differences from the previous frame.
B-frame (Bidirectional) — predicts from both earlier and later frames.
This pattern is what makes inter-frame compression efficient — and what makes such codecs less editable than intra-frame-only ones like ProRes or FFV1.

IPS (In-Plane Switching) — An LCD panel technology common in modern monitors. Offers wider viewing angles and better colour accuracy than the cheaper TN (Twisted Nematic) panels it largely replaced.

IPv6 — The current internet addressing scheme, with far more addresses than the older IPv4, which increasingly lets a device be reached directly without the usual home-router workarounds. One way to reach your own home for off-site backup. See Self-hosting your off-site backup.

IRE — A unit for measuring analog video signal level. Named for the Institute of Radio Engineers (now the IEEE). 100 IRE is reference white, 0 IRE is reference black. NTSC (except in Japan) traditionally has a 7.5 IRE “setup pedestal” — a black-level offset above zero — which is the source of the washed-out blacks you sometimes get on NTSC captures.

Isopropyl alcohol (IPA) — A solvent sometimes added in small amounts to record-cleaning fluid to speed drying. Use with care: it can harm shellac 78s and some labels and finishes, so many archivists avoid it on older or fragile records.

JEDEC — The Joint Electron Device Engineering Council, the standards body for solid-state memory specifications. JEDEC’s published standards include the data-retention specifications that explain why SSDs aren’t suitable for unpowered long-term storage: consumer-grade SSDs are specified to retain data for only around one year at thirty degrees Celsius when unpowered. Enterprise-grade flash has longer retention; in either case, archive use isn’t the design target. The relevant standard for retention is JESD218.

J

Jitter — Tiny variations in the timing of a signal. Time-base errors in VHS playback are a form of jitter; a TBC corrects them.

JPEG — The ubiquitous lossy image format from 1992. Great for sharing and viewing; not great as an archive because each re-save loses a little more detail. Use it for the viewer copy you hand around, not the master file you preserve.

K

Kodachrome — Kodak’s reversal film (slide film) from 1935 to 2009. Unlike Ektachrome, its dyes were added during processing rather than being built into the stock, giving exceptional long-term colour stability. 1950s and 1960s Kodachrome reels often still look excellent today.

L

Lagarith — A lossless video codec, used the same way as HuffYUV: bit-for-bit lossless, so it preserves a capture exactly, at the cost of large files.

LCD (Liquid Crystal Display) — The dominant flat-panel display technology, using liquid crystals modulated by an electric field to control light from a backlight.

ld-decode — The original software-decode project of the family. LaserDisc is an analog video format — the disc is optical, but the video signal on it is FM-modulated analog, not digital. ld-decode captures the raw RF coming off the player’s laser pickup, before the player’s internal demodulator gets to it, and decodes the video in software instead. (LaserDisc’s later PCM audio tracks are digital, and ld-decode handles those as a separate stream.) vhs-decode and cvbs-decode are downstream projects. See: How vhs-decode actually works.

Line TBC — A type of time-base corrector that operates line-by-line on the video signal as it plays. Found inside professional and high-end consumer VCRs like the Panasonic AG-1980 and JVC HR-S9000 series. The most effective TBC for VHS playback.

Linear interpolation — A method of filling in missing or damaged data by drawing a straight line between known good points on either side. In the context of analog video capture, vhs-decode uses linear interpolation between adjacent good lines to replace a line lost to dropout, which is a closer approximation to the missing content than the sample-and-hold method consumer VCRs use.

Lossless — Encoding that allows perfect reconstruction of the original signal. Mathematically lossless codecs (FFV1, FLAC, HuffYUV) decode to bit-identical data. Distinct from visually lossless, which is a weaker claim.

Lossy — Encoding that discards information considered unlikely to be perceived. Most consumer codecs (JPEG, MP3, AAC, H.264, H.265) are lossy.

Loudness war — The long industry practice of mastering music ever louder through heavy compression and limiting, at the cost of dynamic range. A record cut from a gentler master can sound better than its streaming version because of the master, not the medium.

LTO tape — Linear Tape-Open, a magnetic tape format used by institutions for cold archival storage. Capacities range from terabytes to tens of terabytes per cartridge. Drives are expensive, but tapes themselves are cheap per TB.

Macroblock — In modern video codecs (MPEG-2, H.264, H.265), the small square block of pixels — typically 16×16 for luma — that gets processed and compressed as a unit. When you see blocky compression artefacts in low-quality video, you’re looking at macroblock boundaries.

Luminance (Luma) — The brightness component of a video signal, as distinct from chroma. The eye is more sensitive to luminance detail than to chroma detail, which is why chroma subsampling works as well as it does.

M

Macroblocking — Blocky compression artefacts that appear when digital video is encoded at too low a bitrate, breaking smooth areas into visible squares. A risk of poorly encoded MPEG-2 DVDs.

Matrix number — The identifier stamped or hand-etched into a record’s run-out groove, identifying the master and stamper used. Collectors read it to tell an original pressing from a reissue.

M-Disc — An optical disc format claiming archival lifetimes of hundreds to thousands of years, using an inorganic data layer. Real-world testing supports decades of stability; rated lifetimes are unverifiable on those timescales. Useful for the irreplaceable-tier of an archive.

Metal-position tape (Type IV) — The highest-performance cassette formulation, using pure-metal particles for the widest dynamic range. No longer made at the original quality, so sealed examples are collectible.

MFP (Multifunction printer) — A combined printer, scanner, and copier (sometimes plus fax) in a single unit. The scanner part uses a CIS sensor — fine for documents, limited for archival photo work because of the shallow depth of field and lower dynamic range.

MiniDV — A small digital videocassette format (see DV) common in consumer and prosumer camcorders from the late 1990s. Because the recording is already digital, it is best transferred losslessly over FireWire rather than re-captured as analog.

MISRC — An open-hardware capture board designed by Stefan Ölsner for the vhs-decode community. It integrates the clock generation with sample capture in a single device, replacing the more cobbled-together CX-card + Clockgen Mod approach. Considered the longer-term replacement for that legacy combination. See: Capture hardware in 2026.

mkvpropedit — A command-line tool from the MKVToolNix suite that lets you edit the properties of an existing Matroska (MKV) file without re-encoding the streams. Adds or modifies tags, language metadata, track names, and segment information after the file has already been written. The standard tool for embedding archive metadata (tape ID, capture date, deck used, operator notes) into FFV1-in-MKV captures post-encode, without touching the underlying video.

Macrovision — A copy-protection scheme that inserts AGC-confusing pulses into the vertical blanking interval of a video signal. Designed to make a recording VCR misbehave while remaining invisible to a TV. Often misdiagnosed as the cause of rolling pictures on tapes that don’t actually have it. The later ColorStripe variant adds interference in the horizontal sync area on top of the same AGC pulses.

Mosquito noise — Shimmering compression artefacts around sharp edges in digital video, caused by the way MPEG compression handles high-contrast detail. Seen on low-bitrate DVDs.

MPEG-2 — A video codec used by DVD, broadcast video, and HDV. Inter-frame compressed. Considered legacy but well supported.

Mould — Fungal growth on tape, encouraged by sustained high humidity, appearing as white, grey or fuzzy patches with a musty smell. It both damages the tape surface and contaminates any deck that plays it, so a mouldy tape should be dealt with before it goes near a machine. See baking, sticky-shed syndrome.

MP3 — A lossy audio codec from the late 1990s. The dominant consumer audio format for over a decade. Largely superseded by AAC (used by Apple and YouTube) for delivery and FLAC for archival, but still in widespread use because every device ever made plays it.

MTBF (mean time between failures) — A manufacturer’s statistical reliability figure for a drive, in hours. It describes a population’s average failure interval, not the lifespan of any single drive, and says little about a drive left on a shelf.

Mylar — A trade name for polyester (PET) film base, used for both magnetic tape and film. Far more stable than the older acetate base, it resists the breakdown that causes vinegar syndrome.

N

NAND flash — The non-volatile memory that stores data in an SSD or memory card as trapped electric charge. It has finite write endurance and slowly loses its charge, and therefore data, when left unpowered.

New-old-stock (NOS) — Old but unused, often still-sealed, equipment or media sold long after manufacture ceased. Covers unused VCRs, sealed blank tapes and spare parts on the second-hand market.

Noise floor — The lowest level of unwanted signal present in an analog (or sometimes digital) chain. On tape, noise floor includes tape hiss, head-amplifier noise, and other random components.

Normalisation — Adjusting an audio (or video) file’s overall level so its loudest peak sits at a chosen reference, shifting everything up or down by the same amount. Unlike compression it doesn’t change the dynamics — it just moves the whole signal.

NAS (Network-attached storage) — A small, always-on computer built to hold shared drives on your home network, so several devices can read and write the same storage. The natural home for a growing family archive, and the base for replicating an off-site copy to a second box elsewhere. See Self-hosting your off-site backup.

Nitrate film — The earliest flexible film base, used before about 1950, made of cellulose nitrate. It is chemically unstable and genuinely flammable as it decays, so it is a safety hazard as well as a preservation one and is best left to professionals. See cellulose acetate, vinegar syndrome.

NLE (Non-Linear Editor) — Editing software that lets you arrange clips on a timeline and freely change the order. The name distinguishes it from older tape-based editing systems, where you had to record one clip at a time onto a master tape in order. Examples: DaVinci Resolve, Premiere, Final Cut, Vegas, Kdenlive, Avid.

nnedi3 — A neural-network-augmented edge-directed interpolator, used inside QTGMC for spatial gap-filling during deinterlacing. The “AI” component is small and narrow — nnedi3 was trained to predict missing pixels in interlaced video, and works at the per-line / per-block level rather than generating whole images. A foundational tool in the AviSynth and VapourSynth deinterlacing toolchain; an example of AI applied surgically to a narrow signal-processing task.

NTSC — National Television System Committee. The analog colour TV standard used in North America, most of South America, and Japan from the 1950s until digital TV took over in the late 2000s. 525 lines per frame, 29.97 frames per second, colour subcarrier at ~3.58 MHz. Distinct from PAL and SECAM, the European standards.

Nyquist sampling theorem — A foundational principle in signal processing: if you want to capture a signal containing frequencies up to X Hz, you must sample it at more than 2X times per second. Below that rate you lose information you can’t get back; above it, you can perfectly reconstruct the original signal (in theory). The basis of every digital-to-analog and analog-to-digital conversion. Often used to argue that 4:2:2 chroma sampling is enough for VHS — see the chroma article for the practical caveats.

O

Object storage — Cloud storage that holds files as discrete ‘objects’ over an S3-compatible interface, billed per gigabyte stored (and usually per gigabyte retrieved). Cheaper at scale than flat-fee consumer backup, but you bring your own client (rclone, restic and the like) and run it yourself.

Oersted (Oe) — The unit of magnetic coercivity, indicating how strong a field a tape needs to record and how resistant it is to erasure. Higher-coercivity formulations such as chrome and metal hold a stronger signal.

Off-site copy — The copy of your archive kept somewhere other than home, so a fire, theft or flood cannot take the originals and the backup together. The third leg of the 3-2-1 rule, it can live in the cloud, on a NAS at a friend’s house, or on a drive handed to family. See Your off-site backup.

OLED (Organic Light-Emitting Diode) — A self-emissive display technology where each pixel produces its own light. Offers perfect blacks and very high contrast; common in high-end TVs and phones. No backlight needed.

Overscan — The outer edge of a video frame that old CRT TVs cropped off (deliberately, to hide edge instability), so consumer content was never designed to be seen there. Some manufacturers used that margin to embed proprietary data, one reason to capture more than just the visible active lines when preserving a tape. See VBI.

Oxide shedding — The loss of magnetic oxide particles from a decaying tape, leaving rusty dust on heads and guides and degrading the signal. A symptom of binder breakdown and sticky-shed.

PAL — Phase Alternating Line. The European (plus Australia, NZ, parts of Asia and Africa) analog colour TV standard from the late 1960s until digital TV took over. 625 lines per frame, 25 frames per second, colour subcarrier at ~4.43 MHz. Generally considered technically superior to NTSC — higher resolution, more robust colour encoding.

Open format — A file format whose specification is freely available and not encumbered by patents. The right choice for archives because it can be decoded indefinitely without licensing or proprietary tooling. Examples: FLAC, FFV1, MKV.

Optical resolution — A scanner’s actual physical sensor density, measured in DPI. Distinct from the “interpolated” or “enhanced” resolution sometimes advertised, where the scanner takes the optical samples it actually has and uses software to invent values between them. Only the optical part adds real detail.

P

Pack-edge damage — Distortion of a wound tape pack when weight presses on the edges of the reel, as happens when cassettes are stored flat or stacked. Avoided by storing tapes upright, like books.

PCM (Pulse-Code Modulation) — Uncompressed digital audio — the native digital audio format on CD, DAT, and most professional capture pipelines.

P/E cycle (program/erase cycle) — One write-and-erase cycle of a flash memory cell. Each cell tolerates a finite number before it wears out, which is what TBW and DWPD ratings ultimately measure.

Phase-change alloy — The recording layer of a rewritable optical disc (CD-RW, DVD-RW), which a laser switches between crystalline and amorphous states. More limited in re-write life and generally less durable than pressed or write-once discs.

Phono preamp (phono stage) — The stage that lifts a turntable’s tiny cartridge signal to line level and reverses the RIAA equalisation a record was cut with. Without it — a turntable straight into a line input — the sound is thin, quiet and top-heavy. Built into many amplifiers (the “PHONO” input) and into USB turntables. See RIAA equalisation.

Phthalocyanine dye — A light-stable organic dye used in better recordable discs, often gold or silver in appearance. More resistant to fading than the cheaper cyanine dye, and used in archival-grade CD-Rs and DVD-Rs.

Pits and lands — The microscopic bumps (pits) and flat areas (lands) moulded into an optical disc’s surface, on CD, DVD, Blu-ray and LaserDisc alike. A player’s laser reflects differently off a pit than off a land. On CD, DVD and Blu-ray, that pattern of reflected light encodes the disc’s binary data — the ones and zeros a computer actually reads. LaserDisc uses the same pit-and-land structure to store an analogue video signal instead, with no digital data or error correction involved. The reflective layer doing the reflecting is the same aluminium layer vulnerable to oxidation and laser rot.

Plasticiser — An additive that keeps vinyl and some plastics flexible. As it slowly migrates out over decades a record can stiffen, and related migration affects the longevity of other plastic media.

PNG (Portable Network Graphics) — A lossless image format with built-in transparency support. Good for archival image storage if 8-bit or 16-bit per channel is enough for your source. For higher bit depth or floating-point data, use TIFF, DPX or EXR instead.

Polyester base — The stable plastic base, also called PET, used for modern film and for VHS and most later tape. It does not suffer vinegar syndrome and is rated to last centuries, so the base is rarely what fails on a VHS tape; the magnetic coating is. See cellulose acetate, vinegar syndrome.

Pre-emphasis — Boosting certain frequencies before recording, to be reversed on playback (de-emphasis) for a flat result and lower noise. Vinyl’s RIAA curve is the classic example, cutting bass and lifting treble in the groove.

Pressed (replicated) disc — A commercial CD or DVD whose data is physically stamped into metal during manufacture, with no dye to fade. Far more durable than a burned recordable disc, though early pressings can still suffer disc rot.

Pressure pad — The small foam-backed felt pad inside a cassette that holds the tape against the head. The foam degrades with age, muffling sound. Re-padding is routine cassette-restoration work.

Print-through — A faint copy of a loud passage bleeding onto the adjacent layer of wound tape, heard as a pre- or post-echo. Worse on thin tape and after long storage, and reduced by storing tapes tails-out.

Proc-amp — A processing amplifier that adjusts brightness, contrast, chroma saturation, and hue of an analog video signal. Often bundled into frame synchronisers and consumer “TBC” boxes.

Progressive scan — A scanning method that draws every line of a frame in sequence, without interlacing. All modern flat-panel displays are natively progressive.

ProRes — A family of intra-frame codecs developed by Apple, used as a mezzanine codec for editing and grading. Visually lossless at the higher profiles, but not mathematically lossless — ProRes 4444 included.

Provider-held keys — The arrangement where a service encrypts your data but keeps the keys itself, so it can read your files, and so can a rogue employee, a court order, or anyone who steals the keys with the data. The standard tiers of most mainstream clouds work this way; the encryption protects against a stolen disk, not against the company. One of the three ‘keys held by’ categories, alongside zero-knowledge and you encrypting it yourself. See Keeping your family archive out of the wrong hands.

Pulldown (3:2 pulldown / IVTC) — A technique for fitting 24-frames-per-second film into 60-field-per-second interlaced NTSC video, by repeating fields in a 3-2-3-2 pattern (one film frame becomes 3 video fields, the next becomes 2, and so on). When you want to recover the original 24p frames from such material, the process is called inverse telecine (IVTC). Common on NTSC store-bought movies on VHS and DVD.

Pumping — See breathing.

Q

QLED (Quantum-dot Light-Emitting Diode) — An LCD variant using quantum dots in the backlight layer to produce more saturated colours. Not the same as OLED — QLED still has a backlight; OLED doesn’t.

QTGMC (Quasi-Temporal Gentle Motion Compensation) — The community-default deinterlacer in the AviSynth and VapourSynth ecosystems. Processes interlaced video field-by-field with motion compensation and temporal noise reduction, producing progressive output that’s significantly cleaner than simpler deinterlacers like yadif or bwdif. Uses nnedi3 internally for spatial gap-filling. Widely regarded as the most natural-looking deinterlacer for tape-grade analog video.

Quantisation — The step in digitising any signal where continuous values get rounded to the nearest available discrete value. Think of it like rounding 3.7 to 4 because you only have whole numbers to work with. Bit depth determines how many values are available — 8-bit gives 256 steps per channel, 16-bit gives 65,536. The more steps you have, the smaller the rounding error. Quantisation is also what lossy codecs use to throw data away — they map similar values to the same quantised level and save space by storing it once.

R

RAID (and RAIDZ) — A way of combining several drives so the array survives one or more of them failing. RAIDZ is ZFS’s version: RAIDZ1 tolerates one dead drive in a group, RAIDZ2 two. It guards against a failed disk, not against deletion, theft or fire, so it is never a substitute for a separate backup. See Self-hosting your off-site backup.

Ransomware — Malicious software that encrypts your files and demands payment for the key. A particular danger to network storage left reachable from the internet: the DeadBolt and QLocker campaigns locked thousands of consumer NAS drives this way. Snapshots the malware cannot reach, and keeping the box off the open internet, are the defences that matter. See Self-hosting your off-site backup.

RCA — A common consumer audio and video connector, often colour-coded (red and white for audio, yellow for composite video). Unbalanced only.

Real-ESRGAN — An open-source AI image and video upscaler, widely used in the community for super-resolution of compressed or low-quality sources. Built around the ESRGAN architecture (Enhanced Super-Resolution Generative Adversarial Network) with training data tuned for “real-world” degradation rather than clean synthetic test sets. Available as a standalone tool, as a VapourSynth plugin, and through GUIs like ChaiNNer. Better on animation and clean digital content than on tape-grade analog material.

Rec. 601 — The colour space standard for standard-definition video, used by all NTSC and PAL broadcast and VHS material. Slightly different colour primaries and gamma to Rec. 709; untagged Rec. 601 content played back as if it were Rec. 709 produces visible colour shifts.

Rec. 709 — The colour space standard for HD video. Modern playback software assumes Rec. 709 by default, which is why tagging is essential when delivering Rec. 601 content from analog sources.

Record-cleaning machine (RCM) — A device that deep-cleans records, either by applying fluid and vacuuming it off or by ultrasonic bath. Used for grimy records or whole collections, beyond what a brush can do.

Redundancy — Keeping more than one copy so the failure of any single drive, disc or building does not lose the data. The whole point of the 3-2-1 rule and of RAID, though redundancy on its own is not a backup: copies that mirror each other also mirror a deletion or a corruption, which is why version history matters too. See Designing a family archive that survives you — Part I.

Reel-to-reel — An analog audio tape format with two separate physical reels — a feed reel and a take-up reel — loaded onto a deck, rather than enclosed in a cassette shell. Used from the 1940s into the 1980s by both home enthusiasts and professional studios. Wider tape and faster transport than cassette gave significantly better fidelity.

Regular 8 (Standard 8) — An amateur cine film format from 1932 to the mid-1970s. Made by exposing 16mm film on one side, then flipping and exposing the other, then slitting it in two after development. Pre-dates Super 8 and uses smaller frames.

Reissue — A later repressing of a record, made after the original run from new or copied masters. Usually far less valuable than a first pressing of the same title.

Relative humidity (RH) — The moisture in the air as a percentage of the maximum it could hold at that temperature. High RH drives tape binder hydrolysis and film and disc decay; archives keep RH low and, above all, stable.

Replication — Copying data from one machine to another automatically, usually on a schedule. With ZFS it sends only the blocks that changed since last time and produces an exact mirror, which is how a NAS keeps an off-site copy in step with home over the internet. See Self-hosting your off-site backup.

Retensioning — Fast-winding a tape fully end to end to even out the wind and relieve pack stresses before playing or after long storage. A gentle maintenance step for long-stored tapes.

Rewritable disc (CD-RW, DVD-RW) — An optical disc that can be erased and re-recorded, using a phase-change recording layer. Convenient, but generally less durable and less reliable for archiving than write-once or pressed discs.

RF capture — A capture method that records the raw radio-frequency signal coming directly off a VCR’s head, before the deck does any of its own decoding. The basis of vhs-decode and other software-decoded workflows. See: Capture hardware in 2026.

RF tap — A modification to a VCR that pulls the raw radio-frequency signal directly off the head before the deck’s internal electronics decode it. The captured RF is then decoded in software (vhs-decode), producing results that surpass what the VCR itself could output. The vhs-decode wiki maintains a “Tap List” of documented tap points across many decks. See: How vhs-decode actually works.

RGB — Red, Green, Blue. The colour space used by computer displays, image sensors, and most image formats. Each pixel is described by three numbers — how much red, green and blue light it should emit. Contrast with YUV/YCbCr, which separates brightness from colour and is used in video formats.

RIAA equalisation — The standard tone curve applied when cutting a vinyl record — bass reduced, treble boosted so the groove behaves — which playback must exactly reverse to restore a flat response. That reversal is the phono preamp’s job; skip it and the record plays back wrong. See phono preamp.

RIFE (Real-time Intermediate Flow Estimation) — An AI frame-interpolation tool that generates intermediate frames between existing ones, used for frame-rate conversion and motion smoothing. RIFE is the AI tool the analog-video community treats with least controversy — it solves a well-defined problem (given two frames, what would the frame between them look like?) where the authenticity concerns of super-resolution don’t apply in the same way.

Rumble — Low-frequency noise from a turntable’s bearing and motor that leaks into the audio. Unlike clicks and pops it’s very hard to remove cleanly afterwards, so it has to be kept down by the quality of the deck.

Run-out groove — The smooth band of vinyl between the last track and the label, where matrix numbers and stamper marks are etched or stamped. The place collectors look to identify a pressing.

S

Shellac — The brittle resin used to make 78 rpm records before vinyl. Heavier and more fragile than vinyl, and damaged by alcohol-based cleaners, so 78s need gentler handling and cleaning.

S.M.A.R.T. — Self-Monitoring, Analysis and Reporting Technology, a drive’s built-in health reporting. Useful for warning of some impending failures, but not a reliable predictor of all of them, so a drive can fail with a clean report.

Surfactant — A wetting agent in record-cleaning fluid that lowers the surface tension of water so it reaches into the groove and lifts grime. Laboratory surfactants such as Tergitol are used in dedicated recipes.

S-VHS — Super VHS, a higher-resolution variant of VHS with separated luminance and chroma on tape and an S-Video output. The prosumer standard in the late 1980s and 1990s.

S-Video — A connector and signal type that carries luminance and chroma on separate wires within the same connector. Better quality than composite (which combines them on one wire); less than full component video.

S3 (S3-compatible) — The de-facto standard interface for object storage, originally Amazon’s. When a provider says it is ‘S3-compatible’, ordinary backup tools already know how to talk to it, so you can point rclone or a NAS at it the same way you would at Amazon. See Choosing a cloud provider.

Sample-and-hold — The dropout-compensation method used by most consumer VCRs. When the head loses contact with the tape and the signal disappears, the VCR’s dropout compensator pastes the previous good line over the bad one. Simple, fast, and visible — the same content is repeated for as long as the dropout lasts, which produces a smeared band rather than a clean replacement. Contrast with linear interpolation.

Sample rate — The number of audio samples taken per second. CD audio is 44.1 kHz; broadcast and video are typically 48 kHz. Higher sample rates (96/192 kHz) capture frequencies above the human hearing range, useful for processing headroom rather than for the listener.

SCART — A 21-pin European AV connector standard that can carry composite, S-Video or RGB depending on wiring and device support, common on European VCRs, TVs and games consoles instead of separate RCA/S-Video sockets. See RGB.

SDI (Serial Digital Interface) — The professional video signal standard, carrying uncompressed video over a coaxial cable with a BNC connector. Standard in broadcast and post-production for connecting cameras, switchers and capture devices. Variants by bandwidth: SD-SDI (270 Mbit/s), HD-SDI (1.485 Gbit/s), 3G-SDI (3 Gbit/s) and so on.

SECAM — Séquentiel Couleur À Mémoire. The third major analog colour TV standard (alongside NTSC and PAL), used in France, Russia and parts of Africa. Same 625-line, 25 frames-per-second frame structure as PAL, but uses a different colour-encoding method — sending the two colour components on alternating lines rather than together, with the receiver remembering the previous line to fill in the missing one. Less common to encounter in archival work than NTSC or PAL.

Sheet-feed scanner — A scanner with an automatic feeder that pulls one sheet at a time past a stationary sensor — like the document feeder on a small office copier. Dedicated photo sheet-feed scanners (like the Epson FastFoto FF-680W) capture roughly one photo per second, making them the right choice for bulk paper photo digitisation.

Sinc filter — In mathematics and signal processing, the theoretical “ideal” filter for reconstructing a continuous signal from its samples — but only in theory. The sinc filter has infinite length, which is impossible to build in the real world. All real filters are finite-length approximations, which is why real-world digital signal processing is always slightly imperfect compared to what Nyquist’s theorem promises in the ideal case.

Snapshot — A frozen, read-only record of how a filesystem looked at one moment, kept cheaply alongside the live data, which ZFS does natively. Because the snapshot still holds the old version, you can roll back a file that was deleted or scrambled, which is the line between a real backup and a plain sync. See Self-hosting your off-site backup.

SNR (Signal-to-Noise Ratio) — The ratio of clean signal to noise in a chain, usually expressed in decibels. Higher is better. VHS chroma has only about 5 bits of usable SNR; the rest is noise.

SQPB (Super-VHS Quasi PlayBack) — A feature on some standard VHS decks that lets them play S-VHS tapes, at roughly VHS resolution. A plain VHS deck without SQPB can’t properly play an S-VHS recording — the picture breaks up — so it matters when matching a deck to your tapes.

SSD (solid-state drive) — A drive that stores data in flash memory with no moving parts. Fast and robust in daily use, but it loses its charge, and its data, when left unpowered for long stretches, so it is a poor choice for cold archival storage. See flash memory, hard disk drive.

Sticky-shed syndrome — A degradation where the binder holding the oxide to the tape backing absorbs moisture and turns gummy, so the tape squeals, sheds and sticks as it plays. It affects audio (reel-to-reel, cassette) and video tape alike, and is temporarily curable by baking. Distinct from vinegar syndrome, which affects acetate film and tape. See baking, vinegar syndrome.

Stylus — The tiny tip (usually diamond) on a turntable cartridge that rides the record groove. A worn or cheap stylus both sounds poor and physically damages records as it plays, so it’s the part most worth getting right. Closely tied to the cartridge it sits in.

Super 8 — An amateur cine film format introduced by Kodak in 1965, still manufactured today. Smaller sprocket holes than Regular 8, giving about 50% more picture area in the same 8mm width. Cartridge-loaded; far easier to use than Regular 8.

Super-resolution — A class of image and video techniques that take a low-resolution input and produce a higher-resolution output by inferring pixels that weren’t directly captured. Modern super-resolution is typically AI-based, with a trained neural network learning to map low-resolution to high-resolution. The fundamental issue is what’s sometimes called the one-to-many problem: many different high-resolution images can downscale to the same low-resolution one, so the model picks one. For analog tape, where the recording bandwidth is the hard limit on detail, any extra pixels produced by super-resolution are invented rather than recovered.

T

Sync vs backup — A sync mirrors changes both ways, so deleting or corrupting a file at one end carries the damage to the other. A backup keeps earlier versions you can return to. They are not the same, and a service sold as one is often quietly the other; what makes a copy a backup is version history. See the backup providers list.

Tape baking — Gently heating a tape suffering from sticky-shed in a controlled low-temperature oven to drive out absorbed moisture, making it playable long enough to digitise. A one-time rescue, not a cure.

Tape equalisation (NAB, IEC/CCIR) — The record/playback tone curves used by analogue tape machines. NAB is the American standard, IEC (also called CCIR) the European one; a tape sounds tonally right only if the deck uses the same standard it was recorded with. Where that’s unknown, capture flat and correct in software.

Tape hiss — The high-frequency random noise heard on analog magnetic tape. Caused by the discrete nature of the magnetic particles in the tape coating. One specific contributor to the broader noise floor.

Tape speed (ips) — How fast tape passes the heads, in inches per second (commonly 1⅞, 3¾, 7½ or 15 ips). Higher speed gives wider frequency response and a better signal-to-noise ratio — which is why music was recorded fast and long spoken-word slow — and the deck must be set to the speed a tape was recorded at, or the pitch and length come out wrong.

TBC (Time-Base Corrector) — A device that corrects time-base errors in an analog video signal. A true TBC operates line-by-line on the signal; many devices marketed as “TBCs” are actually frame synchronisers with a proc-amp. See: Why a Time Base Corrector matters.

TBC-Tools – The standard post-decode pipeline for the tape (vhs-decode) and cvbs-decode projects, from inspection, adjustment, audio handling to full losless & proxy exported video files and digital replica of the analouge ingest handling stations of old.

TBW (terabytes written) — An SSD endurance rating: the total data that can be written to the drive over its life before the flash wears out. Rarely the limiting factor for a write-once, read-many archive.

Telecine — The process of transferring film to video. Originally a real-time process using a film projector and a video camera; modern telecine is usually frame-by-frame digital scanning.

Teletext — A broadcast text-data service transmitted in the VBI of PAL signals, providing news, programme schedules, subtitles and similar information accessed via the TV remote. Largely defunct as a public service but the signal is still present on archival PAL recordings.

Tergitol — A laboratory non-ionic surfactant, notably Tergitol 15-S-7, used in archival record-cleaning formulas as the wetting agent that lifts grime from the groove.

Test pressing — A handful of copies pressed before a full run, for approval. Usually only five to ten exist, making them a recognised and valuable collectible in their own right.

TIFF (Tagged Image File Format) — A versatile lossless image format. Supports 8-bit, 16-bit and 32-bit per channel, multiple channels, and various compression schemes (including no compression). The standard archive format for scanned photos and film, alongside DPX and PNG.

Time-base error — The variation in line and frame timing of a video signal as it plays back. Causes horizontal jitter and may prevent capture cards from locking. Corrected by a TBC.

Total harmonic distortion (THD) — A measure of the unwanted harmonics a system adds to a signal, as a percentage. Good digital sits around 0.002 to 0.003%; vinyl runs markedly higher, and that added colouration is part of its sound.

Track format (full-, half-, quarter-track) — How the audio tracks are laid across the width of a tape. Full-track uses the whole width (mono); half-track splits it in two; quarter-track packs four, as two stereo pairs recorded in opposite directions (flip the reel to play the second, like a cassette). The deck’s heads must match the tape’s format, or you’ll read the wrong tracks — including a reverse-direction pass playing backwards.

Tracking — The mechanical alignment between a VHS deck’s head drum and the recorded track on tape. Mistracking shows as bands of noise or instability; the tracking knob lets the user fine-tune the alignment for a given tape.

Track width — How much of the tape’s width a single recorded track occupies. A wider track carries more signal relative to the tape’s noise, improving the signal-to-noise ratio — roughly a 3 dB gain for each doubling of track width. It’s the main reason narrower multi-track formats are noisier.

TVL (Television Lines) — The standard measurement of horizontal resolution for analog video. Counts the number of distinct vertical line pairs that can be resolved across a width equal to the picture’s height. Approximate native TVL for common consumer formats: VHS ~240 (NTSC) / ~250 (PAL), S-VHS ~400, Video8 ~240, Hi8 ~400, MiniDV ~500. Broadcast SD sits around 500; LaserDisc CAV around 425. PAL versions of a given consumer tape format tend to sit slightly above their NTSC counterparts because the luma FM carrier is a little higher. TVL is a resolving-power measurement of what the recording format can actually carry, not a count of pixels in a digital file — a correct SD digitisation already captures whatever TVL the tape recorded; sampling at higher rates does not invent more TVL.

Type II (high-bias) cassette — A chrome or cobalt cassette formulation offering lower noise and better treble than standard ferric (Type I) tape, using the deck’s high-bias setting.

U

Two-factor authentication (2FA) — A second proof of identity beyond a password, usually a code from an app or a hardware key, so a stolen or guessed password is not enough to get in on its own. Worth turning on for every account and device that guards an archive. See Keeping your family archive out of the wrong hands.

Ultrasonic cleaning — Deep-cleaning records in a fluid bath agitated by high-frequency sound, whose microscopic cavitation lifts grime from the groove. The most thorough record-cleaning method, suited to collections.

Uncompressed audio — Audio stored with no compression at all — larger file sizes, absolutely lossless. PCM in a WAV or AIFF container is the typical form.

USB-A — The familiar flat, rectangular USB plug found on most computers. In a capture chain it is the far end of many USB capture devices and audio interfaces — it carries the digitised signal to the computer, not the analog one.

USB-C — The smaller, reversible USB connector on newer machines and interfaces. Capable of far more bandwidth than USB-A, though for standard-definition capture USB-A is rarely the bottleneck.

UT Video — A lossless video codec, like HuffYUV and Lagarith — bit-for-bit lossless, large files, a common choice for an archival capture master.

V

v210 — A 10-bit 4:2:2 uncompressed video pixel format used by transfer houses, broadcast archives, and the BBC. Large files, but no codec processing in the chain.

VANC (Vertical Ancillary data) — The digital-video equivalent of the analogue VBI: a reserved region of a DV or SDI signal used to carry metadata such as timecode or captions, alongside rather than inside the visible picture. Sony DV camcorders store closed captions here rather than as analogue line-21 data. See VBI, DV.

VapourSynth — A frame-server framework with a Python scripting interface, used for video-processing pipelines. The modern successor to AviSynth in most post-production workflows. Supports a wide range of filters including QTGMC, neural-network-based filters, and AI tools wrapped as VapourSynth plugins. Most serious vhs-decode post-production chains run through VapourSynth at some stage.

VBI (Vertical Blanking Interval) — The brief period between video fields when the scanning beam returns from the bottom of the screen back to the top, leaving a “blank” interval in the signal. Used for non-picture data: closed captions on Line 21 (NTSC) or teletext (PAL), VITC timecode, copy-protection pulses (Macrovision), and other ancillary signals. Most consumer captures discard the VBI; vhs-decode and similar RF workflows preserve it.

VCR (Videocassette Recorder) — The consumer term for a VHS or S-VHS deck. Largely interchangeable with “deck” but more general.

VHS (Video Home System) — The dominant consumer videocassette format from the late 1970s: half-inch tape, colour-under, standard definition. Its modest bandwidth is why playback quality leans so heavily on the deck and time-base correction. (See also S-VHS, VCR.)

VHS-C — A compact VHS cassette used in camcorders, playable in a full-size VCR via a mechanical adapter. The tape and recording are identical to VHS — only the shell differs — so it digitises exactly like VHS.

vhs-decode — An open-source project for software-decoding raw RF captures from various tape formats. Produces output equivalent to or better than the VCR’s own playback electronics, at the cost of a more involved capture setup. See: How vhs-decode actually works.

VHS recording speeds (SP, LP, EP/SLP) — VHS tapes can be recorded at different speeds to trade quality for runtime:
– SP (Standard Play) — best quality, shortest runtime (2 hours per T-120 tape in NTSC).
– LP (Long Play) — doubles the runtime, at a noticeable quality cost.
– EP / SLP (Extended Play / Super Long Play) — triples the runtime, with significant quality degradation.
Most consumer VCRs can play back all three. The speed a tape was recorded at determines what quality you can capture from it.

Video8 / Hi8 — A family of 8mm camcorder tape formats. Video8 is the baseband 8mm format; Hi8 is its higher-resolution successor (the same step up that S-VHS is from VHS). Both are colour-under and need an Video8, Hi8 or backwards compatible Digital8 camcorder or deck to play or extract FM RF from.

Viewing copy — A delivery encode (typically H.264 or H.265 in MP4) that is easy to play and share. Regenerated from the archival master, never used as a replacement for it.

Vinegar syndrome — The chemical breakdown of acetate-based film (and some tapes), producing acetic acid as a by-product. Affected film smells distinctly of vinegar and becomes brittle and shrunken. Needs professional handling; do not run it through a household projector.

Visually lossless — A codec quality target where compression artefacts are below the threshold of casual perception. ProRes and DNxHR are visually lossless at higher profiles. Distinct from mathematically lossless, which is a stronger guarantee.

VITC (Vertical Interval Timecode) — Timecode information embedded in the VBI of a video signal. Allows precise frame-accurate addressing and synchronisation across multiple devices. Used in broadcast production and preserved by RF captures.

Vorbis comments — A simple key-value metadata format used inside FLAC and Ogg Vorbis files. Arbitrary fields can be embedded directly inside the file, so the metadata travels with the audio. The FLAC equivalent of Matroska tags; for the RF-capture workflow it’s the canonical way to embed capture-chain metadata (deck, date, operator, condition notes) into the RF FLAC file rather than relying on a separate sidecar that can be lost in a move.

vrecord — An open-source automated capture tool from the AMIA (Association of Moving Image Archivists) community. Levrages the Blackmagic or BMD line of PCIe and Thunderblot/USB 3.0 capture devices, providing an industry standard replacement for Media Express with a real-time quality-check monitoring during capture (vectorscope, waveform, audio levels, dropped-frame counter). The preferred open-source workflow for analog video capture but mainly direc to FFV1/FLAC recording over the limited ProRes/V210 options of Media Express, alongside Deck control features for automation and remote capture.

W

WAV (Waveform Audio File Format) — Microsoft’s standard container for uncompressed PCM audio, dating from 1991. Lossless, simple, universally supported. The de facto interchange format for audio.

WinDV — A tiny free Windows utility for capturing DV (digital video) from MiniDV camcorders and similar devices over FireWire. No editing features, just a Start button and a dropped-frame counter. The canonical tool for DV transfer on Windows for over 20 years.

Wow and flutter — Audible pitch wavering from uneven tape or disc speed — slow variations (“wow”) and fast ones (“flutter”). It comes from the mechanics of the deck or turntable and, like rumble, is hard to fix afterwards, so it’s kept down by good, well-maintained playback gear.

X

XLR — A locking professional audio connector with a balanced (signal / ground / inverted-return) 3-pin layout. Standard for microphones, line-level interconnect, and studio gear.

Y

Y/C — A short form for separated luminance and chroma. Also another name for the S-Video signal type.

YCbCr (Y’CbCr) — The digital video colour space, with Y as luminance and Cb/Cr as the two chroma components. Used in all common video codecs that don’t store full RGB.

YPbPr — The analogue component-video signal: luminance (Y) plus two colour-difference signals, carried on three separate cables. The analogue counterpart to digital YCbCr, not RGB, despite the three-cable red/green/blue-coded look. See component video, YCbCr.

YUV — A family of colour spaces used in video that separate brightness (Y) from colour information (U and V). The exact specifications vary — YCbCr is the digital variant used in modern codecs; YUV historically referred to analog component video. Separating brightness from colour is what enables chroma subsampling, which works because the human eye is far more sensitive to brightness detail than colour detail.

Zero-knowledge — A claim that a service cannot read your data because only you hold the key. The plain test is recovery: if the provider can reset your password and hand your account back, it holds a key and the claim fails. Even where it is genuine you are trusting closed software, and it has been broken in the open (MEGA, 2022). For data that truly matters, encrypt it yourself first. See Keeping your family archive out of the wrong hands.

ZFS — A storage filesystem, built into TrueNAS and QNAP’s QuTS hero models, that checksums every block and repairs silent corruption (see bit rot). It also takes cheap point-in-time snapshots and can send them block-for-block to another machine, which is what turns a self-hosted off-site copy into a true, verifiable backup rather than a second pile of files. See Self-hosting your off-site backup.