What is `blocksize` and callback's `frames` param?

I need a little help to understand what is blocksize and callback’s frames param.

At first I thought len(outdata) equals frames, and so it is with play-a-sine-signal example (on my machine), but no with play-a-very-long-sound-file

I read the api section in doc:https://python-sounddevice.readthedocs.io/en/0.3.15/api/streams.html, but I still don’t understand what is the relationship between blocksize, frames, and len(outdata). could you please explain more? Thank you

[hyansuper]

Thank you, that answers my question

[mgeier]

I was surprised for a moment, but then I saw the reason for the confusion:

The play_long_file.py example uses a RawOutputStream. This means that the outdata argument in the callback function is not a NumPy array but a Python buffer object. When you use len() on such an object, you get the size of the data in bytes.

In this case, you should look at stream.samplesize, which is the number of bytes per sample. Since the example uses dtype='float32', you’ll see that the sample size is 4. You’ll also have to check the number of channels (which in your file alert.wav seems to be 1).

The length in bytes of the output buffer is the number of frames times the number of channels times the sample size. This is why in your case len(outdata) is 4 times the number of frames, because the sample size is 4 and the number of channels is 1.

So I was wrong to say that above:

Within the callback function, this is always True:

frames == len(outdata)

This is only true when you use NumPy, i.e. when you use Stream and OutputStream.

When you use RawOutputStream, this is more accurate:

frames * stream.samplesize * stream.channels  == len(outdata)

When you use RawStream, there are pairs of values for samplesize and channels, the second one is the value corresponding to the output:

frames * stream.samplesize[1] * stream.channels[1]  == len(outdata)

Does that make sense?