AIFF Container
Audio Interchange File Format stores sampled audio and supporting information in a chunk-based file structure originally defined by Apple.
An AIFF stores sample descriptions and sound data in separate chunks. Its outer FORM structure identifies the format; COMM describes the audio, and SSND holds its sound data and positioning fields.
| Chunk | Reader use |
|---|---|
| COMM | Obtain channels, sample-frame count, precision, and sample rate |
| SSND | Locate the actual sound samples, accounting for its offset information |
| MARK, where present | Read named positions in the recording |
| Descriptive/instrument chunks | Interpret additional recording or instrument information |
The reader follows sizes through the chunk structure. It does not infer where samples begin solely from the filename or one assumed header length.
Interpret the bytes in the declared order
Original AIFF PCM stores multibyte values in big-endian order: the most significant byte comes first. For a positive 16-bit value of 291, hexadecimal 0123, the stored bytes are 01 followed by 23. Swapping their interpretation produces hexadecimal 2301, or 8,961, a very different sample.
This does not give big-endian audio a sound-quality advantage. It specifies how bytes represent the same numbers. A proper conversion into a little-endian PCM arrangement swaps storage order as required while preserving the decoded values.
For stereo, one sample frame contains a value for each channel. If COMM reports 44,100 sample frames at 44.1 kHz, the audio lasts one second. At 16 bits per channel, the uncompressed sample data occupies 44,100 × 2 × 2 = 176,400 bytes, before container structures.
AIFF-C describes more encodings
AIFF-C extends the description with an encoding identifier, allowing compressed audio and further representations. The extra letter therefore changes what the reader must interpret, rather than simply naming a smaller version of the same original PCM layout.
To compare an AIFF master with a WAV listening copy in an Emby collection, compare decoded format, rate, channel layout, and complete duration. A conversion preserving those samples can still write different bytes and different tags because the destination structures differ.
An accurate sample comparison begins after each container and byte arrangement has been interpreted. Comparing the raw file bytes would conflate the sample values with their storage order, headers, and descriptive chunks.