selecting a specific microphone leads to issues

[Souvik Pal]

I am using a USB microphone that supports 44100 and 4800 HZ as sampling rate.

When I use

Get microphone stream

FORMAT = pyaudio.paInt16
CHANNELS = 1

RATE = 16000

RATE = 44100
CHUNK = 4096
audio = pyaudio.PyAudio()
mic_stream = audio.open(format=FORMAT, channels=CHANNELS, rate=RATE, input=True, frames_per_buffer=CHUNK)

The code works fine, but when I specify the specific mic,

MIC_INDEX = 0

Get microphone stream

FORMAT = pyaudio.paInt16
CHANNELS = 1

RATE = 16000

RATE = 44100
CHUNK = 4096
audio = pyaudio.PyAudio()

mic_stream = audio.open(format=FORMAT, channels=CHANNELS, rate=RATE, input=True, frames_per_buffer=CHUNK)

mic_stream = audio.open(format=FORMAT, channels=CHANNELS, rate=RATE, input=True, frames_per_buffer=CHUNK, input_device_index=MIC_INDEX)

I get the following error

INFO: Created TensorFlow Lite XNNPACK delegate for CPU.

        Model Name         | Score | Wakeword Status
        --------------------------------------
        screen_off         | 0.000 | --                    #####################################
        play_my_messages   | 0.000 | --
        take_a_selfie      | 0.000 | --                    #####################################
        close_door         | 0.000 | --
        window_mode        | 0.000 | --
        selfie_mode        | 0.000 | --
        open_door          | 0.000 | --

Traceback (most recent call last):
File "/home/souvik/localapp/bdd/engine/agent1.py", line 96, in
audio = np.frombuffer(mic_stream.read(CHUNK), dtype=np.int16)
File "/home/souvik/localapp/bdd/localenv/lib/python3.10/site-packages/pyaudio/init.py", line 570, in read
return pa.read_stream(self._stream, num_frames,
OSError: [Errno -9981] Input overflowed

Any pointers on why is this happenning?

[dscripka]

pyaudio can be difficult to setup when you have multiple audio devices. I would recommend testing the script separately without openWakeWord to ensure that the pyaudio configuration is working correctly.

Also, note the openWakeWord only supports 16khz audio.