How to convert an audio stream from PYAudio into the right format to pass to noisereduce? about noisereduce HOT 5 CLOSED

I was able to make some progress but all the values in the output of the noisereduce are 0

mic_data = await audio_queue.get()

numpydata = np.frombuffer(mic_data, np.int16)

reduced_noise = nr.reduce_noise(
  y=numpydata, 
  sr=SAMPLE_RATE,
  stationary=True,
  freq_mask_smooth_hz=2000,
  time_mask_smooth_ms=75,
  n_std_thresh_stationary=2
)
noise_mic_data = reduced_noise.tobytes()

from noisereduce.

Did you see #44 already?
If not the approach + snippets there might help you

from noisereduce.

I figured out the issue actually

I was trying to send smaller chunks of audio 20ms at a time and up to 200ms at a time.

Once I switched to 1sec audio chunks the above code started working for me

from noisereduce.

Closing as the above code does the job once frames per buffer and the audio stream was set to 1 seconds (16k in my case since the audio is 16Khz)

from noisereduce.

how to fix？

from noisereduce.