Hello all,

I am researching existing wave formats with a view to storing fairly large epochs of recorded waveforms from various sensors - sampled usually at around audio frequencies. I was looking for a format specification for WavPack eg, how many channels supported, sampling frequencies supported, how to embed metadata etc? Does a formal format specification exists anywhere?

For anyone interested my requirements include:

Multiple Signals (Channels) - probably up to 256.
Each channel may have independant sample rates.
Channels may be added/removed during recording.
Substantial time-stamped metadata may be stored during recording.
Data written is inviolate, ie no going back and changing anything earlier in the stream.

Thanks for any help.

I'm having the feeling that you'll have a hard time finding a codec that supports independent sampling rates for the channels, as it is almost never needed in audio.

That said, a codec could support that without too much problems. The only question would be where to store the data (the .wav format cannot do so)

The closest I know to a formal format specification is here

http://www.wavpack.com/file_format.txt

I know that WavPack could technically support upto 255 channels (it allocates one byte to hold the number of channels) but not sure if this is the maximum number in reality.
While WavPack has 15 predefined supported sample rates, it is in fact possible to define/use your own sample rate.
It does not support independent sampling rates for each channel, sorry.

Thanks all for the feedback.