https://en.wikipedia.org/wiki/Advanced_Audio_Coding
Free to distribute, but not develop with.
https://en.wikipedia.org/wiki/Vorbis
https://en.wikipedia.org/wiki/Speex
Vorbis predecessor.
https://en.wikipedia.org/wiki/MP3
MPEG-1 or MPEG-2 Audio Layer III.
https://en.wikipedia.org/wiki/MPEG-1_Audio_Layer_II
More popular than MP3 for broadcasting. TODO why?
MPEG-1 Audio Layer II or MPEG-2 Audio Layer II.
Synthetic generation of pure tone + to PCM FFmpeg C API decode example: https://github.com/FFmpeg/FFmpeg/blob/1ec7a703806049265991723a8826bd61555edef4/doc/examples/decoding_encoding.c
No metadata, just bytes.
https://en.wikipedia.org/wiki/Raw_audio_format
This leaves the format unspecified: you should at least specify:
- sampling rate
- number of bits per byte
- endianess
- number of channels
- type of data (signed, unsigned, floating-point, mu-law, a-law, ...)
ffplay from FFmpeg and play from SoX can play it.
- http://stackoverflow.com/questions/20314739/how-to-play-pcm-sound-file-in-ubuntu
- http://superuser.com/questions/76665/playing-a-pcm-file-on-a-unix-system
IBM, Windows. Widely Linux implemented.
https://en.wikipedia.org/wiki/WAV
Is a RIFF format https://en.wikipedia.org/wiki/Resource_Interchange_File_Format
Contains metadata, so unlike RAW it contains enough data for players to play it properly.
TODO? vs raw?
https://en.wikipedia.org/wiki/Pulse-code_modulation
Decoded from MP2: https://github.com/FFmpeg/FFmpeg/blob/1ec7a703806049265991723a8826bd61555edef4/doc/examples/decoding_encoding.c
Apple, derived from IFF (Acorn).
https://en.wikipedia.org/wiki/Audio_Interchange_File_Format
44100 Hz is a common sampling frequency: https://en.wikipedia.org/wiki/44,100_Hz
Humans hear between 20 and 20k Hz.
https://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem implies that at least 40k is needed on a perfect system.
TODO what is this?