File format, media and binary structure glossary

Plain-English technical definitions for 14 foundational concepts that come up when testing file uploads, building binary parsers, or debugging audio/video transcoding pipelines.

Core terms and binary definitions

Byte Order Mark (BOM)
A Unicode code point (U+FEFF) placed at the very start of a text stream (encoded as EF BB BF in UTF-8, FE FF in UTF-16BE, or FF FE in UTF-16LE). While optional in UTF-8, prepending EF BB BF to a CSV file tells Microsoft Excel on Windows to interpret the file as UTF-8 rather than a legacy ANSI codepage.
Central Directory (ZIP)
The master index table stored at the end of a ZIP archive (and inside DOCX, XLSX, PPTX, ODT, and EPUB containers). Because the Central Directory is written at the end of the file after all compressed entries, a truncated ZIP download fails when an unzipper seeks to the End of Central Directory (EOCD, signature 50 4B 05 06).
Chroma Subsampling (4:4:4, 4:2:2, 4:2:0)
The practice of encoding color-difference (Cb, Cr) channels at a lower spatial resolution than luminance (Y) because the human eye is more sensitive to brightness detail than color detail. 4:2:0 halves both horizontal and vertical chroma resolution (requiring even pixel dimensions), whereas 4:4:4 preserves full color resolution on sharp UI text.
Codec vs. Container
A container format (such as MP4, MKV, WebM, AVI, Ogg, or WAV) defines the outer file structure, metadata boxes, and how video, audio, and subtitle tracks are interleaved. A codec (such as H.264, HEVC, VP9, AV1, AAC, Opus, or FLAC) is the mathematical algorithm that compresses and decompresses the media stream inside that container.
Constant vs. Variable Bitrate (CBR vs. VBR)
Constant Bitrate (CBR) allocates the exact same number of bits per second regardless of scene or audio complexity, making file size linearly proportional to duration. Variable Bitrate (VBR) allocates more bits to complex frames and fewer bits to silence or static scenes, achieving higher quality per byte.
Constant vs. Variable Frame Rate (CFR vs. VFR)
Constant Frame Rate (CFR) video displays frames at a fixed interval (such as 24.00 or 29.97 fps). Variable Frame Rate (VFR) video—common in smartphone recordings and screen captures—varies the timestamp delta between frames, which can cause audio-sync drift in naive video editors.
Cross-Reference Table (PDF xref)
The byte-offset lookup table placed near the end of a PDF file right before the trailer dictionary and %%EOF marker. A PDF reader reads the startxref offset at the end of the file first, jumps to the xref table, and uses those byte offsets to locate individual numbered objects (pages, fonts, images) without scanning the entire file.
Endianness (Big-Endian vs. Little-Endian)
The order in which multi-byte integers are stored in memory or on disk. Big-endian ('network byte order', used in PNG, JPEG, AIFF, SQLite headers, and Java .class) stores the most significant byte first; little-endian (used in WAV, AVI, WebP, BMP, ZIP, and WebAssembly) stores the least significant byte first.
Lossless vs. Lossy Compression
Lossless compression (PNG, WebP Lossless, FLAC, ALAC, WAV, ZIP, Parquet) reconstructs the exact original bits upon decompression. Lossy compression (JPEG, MP3, AAC, Opus, H.264) permanently discards perceptually masked high-frequency detail to achieve 5× to 20× smaller files.
Magic Bytes (File Signature)
Fixed byte sequences at known offsets (usually offset 0, 4, or 8) that identify a file's binary format independently of its filename extension—for example 25 50 44 46 2D ('%PDF-') for PDF, 89 50 4E 47 for PNG, or 50 41 52 31 ('PAR1') for Apache Parquet.
MIME Type (Media Type)
An IANA-standardized two-part identifier (type/subtype, such as application/pdf, image/avif, or text/csv) transmitted in the HTTP Content-Type header to tell browsers and APIs how to interpret a payload without guessing from the URL.
Sample Rate and Bit Depth (PCM Audio)
Sample rate (measured in Hz, such as 44,100 Hz for CD audio or 48,000 Hz for video/Opus) is the number of amplitude snapshots captured per second per channel. Bit depth (such as 16-bit integer or 24-bit/32-bit float) defines the dynamic range and quantization precision of each sample.
SHA-256 Checksum
A 256-bit (32-byte, rendered as 64 hexadecimal characters) cryptographic hash function from the NIST SHA-2 family. Comparing the SHA-256 digest of a downloaded file against its published hash proves that not a single bit was altered or truncated in transit.
Solid Archive Compression
An archive layout (used by default in .7z and .tar.xz/.tar.zst, unlike .zip) where multiple input files are concatenated into a single byte stream before compression so the dictionary can exploit redundancy across similar files.

Apply these concepts with our diagnostic tools