libburn/doc/waveformat.txt

85 lines
3.4 KiB
Plaintext

Sound extraction for CD-DA burning from .WAV audio file format
Using information and text snippets
from https://ccrma.stanford.edu/courses/422/projects/WaveFormat/
in may 2013. The link is now dead. An apparent copy of the page
is 2017 at: http://soundfile.sapp.org/doc/WaveFormat/
from https://en.wikipedia.org/wiki/WAV
For libburnia-project.org by Thomas Schmitt <scdbackup@gmx.net>
December 2017
The WAVE file format is an application of the Microsoft RIFF container format
for multimedia files. A RIFF file consists of Chunks which contain Subchunks.
The Chunks form a linked list within the file, the Subchunks form a linked
list inside their Chunk.
All numbers are stored in little-endian byte order.
A .WAV file consists at least of one Chunk with id "RIFF", which contains
one Subchunk with id "fmt " and one with id "data":
Offset Size Name Description
0 4 ChunkID Contains the letters "RIFF"
4 4 ChunkSize The size of the rest of the chunk following
this field. I.e. the two fields ChunkID and
ChunkSize are not included in this count.
8 4 Format Contains the letters "WAVE"
The "fmt " subchunk describes the sound data's format:
Offset Size Name Description
0 4 Subchunk1ID Contains the letters "fmt "
4 4 Subchunk1Size The size of the rest of the Subchunk following
this field. I.e. Subchunk1ID and Subchunk1Size
are not included in this count.
8 2 AudioFormat PCM = 1 (i.e. Linear quantization)
Values other than 1 indicate some
form of compression.
10 2 NumChannels Mono = 1, Stereo = 2, etc.
12 4 SampleRate 8000, 44100, etc.
16 4 ByteRate == SampleRate * NumChannels * BitsPerSample/8
20 2 BlockAlign == NumChannels * BitsPerSample/8
The number of bytes for one sample including
all channels.
22 2 BitsPerSample 8 bits = 8, 16 bits = 16, etc.
More data may follow in this Subchunk if AudioFormat is not PCM.
The "data" subchunk contains the size of the data and the actual sound:
Offset Size Name Description
0 4 Subchunk2ID Contains the letters "data"
4 4 Subchunk2Size == NumSamples * NumChannels * BitsPerSample/8
The number of audio data bytes.
8 * Data The audio data bytes.
CD-DA prescribes these "fmt " parameters:
AudioFormat == 1
SampleRate == 44100
BitsPerSample == 16
NumChannels == 2 (stereo)
(little-endian byte order)
If matching parameters are given in the .WAV file, one can directly use the
data bytes of Subchunk "data" as payload for burning a CD-DA track.
Above simple form can be expanded by other Chunks or Subchunks of Chunk "RIFF".
A .wav file appeared which beared a Subchunk "LIST" inside Chunk "RIFF".
Wikipedia mentions Chunks "INFO", "CSET", "JUNK", "PAD ".
Therefore one should expect such Chunks before Chunk "RIFF" and Subchunks
other than "fmt " and "data" inside the "RIFF" Chunk.
Multiple Chunks "RIFF" and Subchunks "fmt " or "data" per file have not been
seen yet. They would make extraction more cumbersome.