rajivpoddar / logmmse Goto Github PK

After processing, some of the noise at the bottom of the waveform is not continuous, and there is a problem of up and down translation, resulting in noise

transcribe-meeting | noise_threshold)transcribe-meeting | File "/usr/local/lib/python3.7/site-packages/logmmse/base.py", line 19, in mono_logmmse
transcribe-meeting | _output, saved_params = _logmmse(signal, fs, initial_noise, window_size, noise_threshold, saved_params)
transcribe-meeting | File "/usr/local/lib/python3.7/site-packages/logmmse/logmmse.py", line 29, in logmmse
transcribe-meeting | noise_mean = noise_mean + np.absolute(np.fft.fft(win * x[j:j + Slen], nFFT, axis=0))
transcribe-meeting | ValueError: operands could not be broadcast together with shapes (882,) (822,)
transcribe-meeting exited with code 1

I am trying to Noise Reduce a wav file using logmmse and I receive the mentioned exception

The noise_threshold is overwritten in the final function (namely eta argument of logmmse) , so it don't work as expected

I am getting this error while executing the logmmse.py file.
Command: python logmmse.py input.wav output.wav

UserWarning: Could not import alsa backend; most probably, you did not have alsa headers when building audiolab
warnings.warn("Could not import alsa backend; most probably, "
Traceback (most recent call last):
File "logmmse.py", line 109, in
output, saved_params = logmmse(signal, fs, args.initial_noise, args.window_size, args.noise_threshold, saved_params)
File "logmmse.py", line 28, in logmmse
x_old = np.zeros(len1)
TypeError: 'float' object cannot be interpreted as an index

Please let me know if there is any fix for this. Thank you.