File format glossary

An A-Z reference of file format and conversion terminology — codecs, containers, lossy vs lossless, transparency, OCR, and the rest of the vocabulary you encounter when working with files.

A

AAC: Advanced Audio Coding — a lossy audio compression format that succeeded MP3. AAC sounds slightly better than MP3 at the same bitrate and is the default audio format for the Apple ecosystem (iTunes, Apple Music, video soundtracks). Files typically have the .m4a or .aac extension.
Alpha channel: An additional data channel in some image formats that stores transparency information for each pixel. PNG, WebP, AVIF, GIF, TIFF, and SVG all support alpha channels; JPEG does not. An image with an alpha channel can have soft, anti-aliased edges or partially-transparent regions.
AVIF: AV1 Image File Format — a modern image format derived from the AV1 video codec. AVIF compresses 30-50% smaller than WebP and 50%+ smaller than JPEG at comparable visual quality, making it one of the most efficient image formats currently in use. Browser support is universal in modern browsers (Chrome since 2020, Firefox since 2021, Safari since 2022/2023).

B

Bitrate: The amount of data used per second to encode an audio or video stream. Higher bitrates mean better quality and larger files. For audio, common bitrates are 128 kbps (acceptable for voice), 192 kbps (sweet spot for music), 256 kbps (high quality), and 320 kbps (maximum for MP3/AAC). For video, bitrate depends on resolution — 1080p typically needs 2-5 Mbps for streaming quality.

C

Codec: Short for “coder-decoder” — the algorithm used to compress and decompress audio or video data. Common video codecs include H.264, H.265 (HEVC), VP9, and AV1. Common audio codecs include MP3, AAC, Opus, and FLAC. The codec is distinct from the container format — for example, an MP4 container can hold H.264 or H.265 video.
Container format: A file format that wraps one or more streams of media data along with metadata. MP4, MKV, MOV, and WebM are video containers; M4A and OGG are audio containers; ICO is an image container that holds multiple resolutions. The container determines the file extension; the codec inside the container determines the actual encoding of the media.

D

DRM: Digital Rights Management — encryption applied to media files to control how they can be used. DRM-protected ebooks, music, and video files cannot be freely converted because the encryption prevents reading the underlying content. Removing DRM is technically possible for some files but is often legally restricted depending on jurisdiction and the file's terms of use.

E

EXIF: Exchangeable Image File Format — metadata embedded in image files that describes camera settings, lens information, GPS coordinates, timestamps, and other contextual data. EXIF transfers cleanly between most image formats during conversion; some Apple-specific or proprietary metadata fields don't translate.
Embedded font: A font file included inside a document so that the document renders correctly even on devices that don't have the font installed. PDF, EPUB, and PPTX all support embedded fonts. Embedding adds to file size but ensures consistent visual presentation.

F

FLAC: Free Lossless Audio Codec — a lossless audio compression format that produces files about 40-60% the size of equivalent uncompressed WAV. Bit-identical to the original PCM audio when decompressed, making it ideal for archival and audiophile use.

G

Generation loss: Cumulative quality degradation that occurs when a lossy file is repeatedly re-encoded. Each encode pass discards more data; doing it five or six times produces obviously degraded output. The fix is to keep a lossless master and only encode to lossy formats once at the moment of distribution.
GLB / glTF: GL Transmission Format — a modern, web-optimised 3D model format from the Khronos Group. GLTF is the JSON-with-external-assets variant; GLB is the single-file binary variant. Both support PBR materials, animations, and skeletal rigs. Often described as “the JPEG of 3D”.

H

HEIC: High Efficiency Image Coding — Apple's default photo format since iOS 11, based on the HEVC video codec. Compresses about 2× more efficiently than JPEG at the same quality. Limited support outside the Apple ecosystem makes conversion to JPEG common.

I

ICO: A Windows icon container format that holds multiple resolutions of the same icon (typically 16×16, 32×32, 48×48, and 256×256) in a single file. Used for browser favicons and Windows desktop icons. Modern web favicons can also be PNG, but ICO remains the most compatible single-file option.

J

JPEG / JPG: Joint Photographic Experts Group — the dominant lossy image format since 1992. Universally supported across every device, browser, and platform. Optimised for photographic content; produces visible artifacts on graphics with sharp edges or text. Quality is typically expressed on a 1-100 scale; 85+ is web-suitable, 92+ is essentially indistinguishable from the source.

L

Lossless compression: Compression that preserves every bit of the original data — the decompressed file is bit-identical to the source. PNG, FLAC, ZIP, and TIFF (with LZW) are all lossless formats. Compression ratios are smaller than lossy formats but the reconstruction is perfect.
Lossy compression: Compression that discards data the encoder believes the human eye or ear won't notice. JPEG, MP3, AAC, H.264, and most video codecs are lossy. File sizes are 10-100× smaller than equivalent lossless formats, but quality degrades with each re-encode pass.

M

Manifold geometry: In 3D modelling, a mesh that forms a closed solid with no holes, gaps, or self-intersecting faces. Manifold geometry is required for 3D printing — non-manifold meshes fail to slice or print incorrectly. Most 3D modelling tools have a “mesh repair” option that fixes common manifold issues automatically.
MOBI: Amazon Kindle's older ebook format, derived from the Mobipocket reader format. Supported by every Kindle device ever made. The newer AZW3 (KF8) format is preferred on modern Kindles, while MOBI remains compatible with older hardware.

O

OCR: Optical Character Recognition — the process of identifying machine-readable text from an image of text. OCR is the prerequisite for converting scanned PDFs (which contain only pictures of pages) into editable text formats like DOCX. Modern OCR is highly accurate (99%+) for clean printed text.

P

PBR: Physically-Based Rendering — a modern approach to computer graphics that simulates how light interacts with surfaces using physically-realistic material properties (metallic, roughness, normal, occlusion). Modern 3D formats like glTF use PBR materials; older formats like OBJ use simpler material models.
PNG: Portable Network Graphics — a lossless image format with full alpha-channel transparency support. Ideal for graphics with sharp edges, screenshots, logos, and any image where lossless quality matters. Larger files than JPEG for photographic content.

R

Resampling: The process of changing an image's resolution. Downsampling (reducing resolution) typically uses Lanczos or bicubic algorithms to preserve edge detail. Upsampling (increasing resolution) cannot recover detail that wasn't present in the source — modern AI-based upsamplers can intelligently interpolate, but conventional resampling produces softer results when scaling up.
RGB / CMYK: RGB (Red Green Blue) is the additive colour model used for screens — colours are produced by mixing light. CMYK (Cyan Magenta Yellow Black) is the subtractive model used for print — colours are produced by mixing inks. Converting between the two can shift colour values noticeably; designs intended for print should be authored in CMYK from the start.

S

Sample rate: The number of audio samples captured per second, measured in Hz or kHz. CD-quality audio is 44.1 kHz; professional audio is often 48 kHz; higher rates (96 kHz, 192 kHz) are used for archival or high-end production. Sample rate affects the highest frequency the audio can represent (Nyquist theorem: max frequency is half the sample rate).
SRT: SubRip Text — the most universally supported subtitle format. A plain text file with numbered cues, timestamp ranges, and the corresponding subtitle text. Read by virtually every video player. The fallback format when you don't know what your destination supports.
STL: STereoLithography — the universal format for 3D printing. Stores triangulated surface geometry only, with no colour, texture, or material information. Required to be manifold (closed solid) for printing.
SVG: Scalable Vector Graphics — the web standard for vector images. Stores artwork as mathematical descriptions of shapes and paths rather than pixels, allowing infinite scaling without quality loss. The right format for logos, icons, and illustrations on the web.

T

Transcoding: Converting media from one codec to another, typically requiring a full decode-and-re-encode cycle. Each transcode introduces some quality loss for lossy codecs. The fix when possible is to remux (change container without changing codec), which is lossless.
Transparency: The ability of an image format to encode partially-transparent or fully-transparent pixels. PNG, WebP, AVIF, and SVG support partial transparency (alpha channel). GIF supports binary transparency only (each pixel is either fully transparent or fully opaque). JPEG does not support transparency at all.

U

UTF-8: Universal character encoding that can represent every character in every language on Earth, plus emoji and symbols. UTF-8 is the dominant encoding on the modern web and the recommended encoding for any text file. Non-UTF-8 encodings (Windows-1252, Latin-1, Shift-JIS, GB18030) appear in older files and can cause character corruption when read by tools that assume UTF-8.

V

VTT (WebVTT): Web Video Text Tracks — the modern web standard for subtitles. Designed for HTML5 video and used by YouTube, Netflix, and most streaming platforms. Extends the simpler SRT format with optional positioning, alignment, and styling features.

W

WAV: Waveform Audio File Format — Microsoft's uncompressed audio container, the standard for professional audio production. Files are large (about 10 MB per minute of stereo CD-quality audio) but contain perfect, uncompressed sample data with no compression artifacts.
WebP: Google's image format from 2010. Supports both lossy and lossless modes plus full alpha transparency. Compresses 25-35% smaller than JPEG at the same visual quality. Universal browser support in 2026.
WOFF / WOFF2: Web Open Font Format — compressed wrappers around TTF or OTF fonts, optimised for fast web delivery. WOFF2 (using Brotli compression) is about 30% smaller than WOFF and is the preferred format for web font deployment in modern browsers.

Y

YAML: YAML Ain't Markup Language — a human-friendly data serialization format used heavily for configuration files (Kubernetes, GitHub Actions, Docker Compose, Ansible). Supports comments, references, and indentation-based structure. The same data tree as JSON but more readable for humans.

Want a term added?

This glossary is curated, not exhaustive. If there's a term you encountered in your conversion workflow that we haven't defined, let us know.

Suggest a term