trouble with record start and record stop about opensesame-plugin-audio_low_latency

[sflavier]

Hi.
I test the new release, 8.2.0, and the same problems occurs.

it looks like, the buffer from the sounddevice need to be converted before storing it in memory or directly in the file by adding :
data =numpy.frombuffer(data[0])

in file audio_low_latency_record_start.py, line 290

[dev-jam]

Hi Sébastien,

Nice to hear you are using this plugin. I was still fixing the sounddevice and pyaudio code. On linux it did work but on Windows it was really buggy. Most of the time we use pyalsaaudio on debian which works pretty stable and gives a 1 ms onset latency. I now found out I was mixing blocking mode code with callback mode code for sounddevice and pyaudio. So I first fixed that part.

I was still looking for converting the numpy array data to byte data and your numpy line seems to work. I did not test it with a real mic yet but the wave file seems consistent now. Thanks!

I am also not really experienced yet with accepting and merging commits. Your code needed an extra if statement so it does not convert the pyaudio data. For now I am gonna merge them manually if that is ok with you and put a reference to you in the commit. And also merge this in the background record start item.

Next on the agenda is to test for stability and look into the file pool object for saving the wave files. I only used this file pool object for opening files, so it follows the default opensesame search procedure for finding the file. And I also find it curious that for playback with sounddevice I did not have to convert from byte data to numpy data, also gonna check that out.

Btw nice to see that it is also now working on the mac, never tested that before.

Cheers!
Bob

[sflavier]

Thanks for the update !!