search

csteinmetz1 / pymixconsole

Headless multitrack mixing console in Python
https://csteinmetz1.github.io/pymixconsole
MIT License
116 stars 13 forks source link

demo.ipynb: several audio (output) clips are silent? #5

Open drscotthawley opened 3 years ago

drscotthawley commented 3 years ago

Christian, this is wonderful. I look forward to learning how we can train differentiable audio effects using your system.

For now I'm just going through the demo notebook, and after the text that reads

"We will apply our processing on the guitar signal from before in the exact same way, except this time we will call the process() method of the channel. Before we apply the processing with the default settings, let's change the equaliser and compressor (by default they will do no processing)."

...there are two code blocks, followed by two audio samples. The second audio clip is silent for me. Is that to be expected? I'm also noticing silence for the second clips in the pairs of audio clips accompanying the Panner, the Delay, and the Reverb.

Perhaps I should mention, I see the following warning somewhat regularly; not sure what its consequences are:

> /home/shawley/env_pymix/lib/python3.8/site-packages/pymixconsole/processors/convreverb.py:88: WavFileWarning: Chunk (non-data) not understood, skipping it.
  sr, h = wavfile.read(filename)   # load the audio file for correct impulse response
drscotthawley commented 3 years ago

AH! I see, I should be trying out your compressor example in auraloss anyway! I'll switch over to that for a while...

csteinmetz1 commented 3 years ago

Hey Scott. This works for me when loading the notebook into Google colab. I have noticed inconsistent behavior with how the audio gets rendered in the notebook when using colab vs. local. I am not sure of the exact cause. I would try loading this notebook in colab though to see if you get the same results.

The warning message that you see does show for me as well, but that should not impact the results. Thanks for checking out the repo.