Unable to run example given on README [sounddevice.PortAudioError]

[ankurdhuriya]

>>> from diart import OnlineSpeakerDiarization
>>> from diart.sources import MicrophoneAudioSource
>>> from diart.inference import RealTimeInference
>>> from diart.sinks import RTTMWriter
>>> pipeline = OnlineSpeakerDiarization()
>>> pipeline.config
<diart.blocks.diarization.PipelineConfig object at 0x16f5909d0>
>>> pipeline.config.sample_rate
16000
>>> pipeline = OnlineSpeakerDiarization()
KeyboardInterrupt
>>> mic = MicrophoneAudioSource(pipeline.config.sample_rate)
||PaMacCore (AUHAL)|| AUHAL component not found.Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/ankurdhuriya/opt/miniconda3/envs/diart/lib/python3.8/site-packages/diart/sources.py", line 125, in __init__
    self._mic_stream = sd.InputStream(
  File "/Users/ankurdhuriya/opt/miniconda3/envs/diart/lib/python3.8/site-packages/sounddevice.py", line 1421, in __init__
    _StreamBase.__init__(self, kind='input', wrap_callback='array',
  File "/Users/ankurdhuriya/opt/miniconda3/envs/diart/lib/python3.8/site-packages/sounddevice.py", line 898, in __init__
    _check(_lib.Pa_OpenStream(self._ptr, iparameters, oparameters,
  File "/Users/ankurdhuriya/opt/miniconda3/envs/diart/lib/python3.8/site-packages/sounddevice.py", line 2745, in _check
    raise PortAudioError(errormsg, err, hosterror_info)
sounddevice.PortAudioError: <exception str() failed>

portaudio==19.7.0
sounddevice==0.4.5

Platform, Operating System, and Programming Language

Platform : Apple M2
OS : macOS Ventura 13.1
Programming Language : Python 3.8.15

[juanmc2005]

Hi @ankurdhuriya,

I don't have an apple device to verify it, but I suspect this may be related to issue #99.
Could you try modifying MicrophoneAudioSource to manually set the correct device ID ?

You can view the available devices with python -m sounddevice

[juanmc2005]

I think I may have a temporary solution to this. Diart provides a TorchStreamAudioSource that you could use to wrap a torchaudio.io.StreamReader linked to the microphone as shown here:

from torchaudio.io import StreamReader
from diart import OnlineSpeakerDiarization
from diart.sources import TorchStreamAudioSource
from diart.inference import RealTimeInference
from diart.sinks import RTTMWriter

pipeline = OnlineSpeakerDiarization()
streamer = StreamReader("0:0", format="avfoundation")
mic = TorchStreamAudioSource(
    uri="live_recording",
    sample_rate=pipeline.config.sample_rate,
    streamer=streamer,
    stream_index=0,
)
inference = RealTimeInference(pipeline, mic, do_plot=True)
inference.attach_observers(RTTMWriter(mic.uri, "/output/file.rttm"))
prediction = inference()

This should automatically grab the Mac's default microphone.
Ideally I would like to have a single class for any device, but a new AppleMicrophoneAudioSource could be a nice temporary solution to quickly get started with the default device.
I would gladly merge a PR with this feature if you're interested in contributing.

[juanmc2005]

This might also be related:

• MacOS BigSur: ||PaMacCore (AUHAL)|| Error on line 1316: err='-66748', msg=Unknown Error spatialaudio/python-sounddevice#299
• https://stackoverflow.com/questions/63543101/error-trying-to-record-audio-using-python-mac

[juanmc2005]

Implemented in #182. Will appear in next release (v0.8).

Feel free to reopen if needed