Voice analysis functions#2689
Conversation
|
@pplantinga let me know if you are satisfied with this PR, it looks good to me. Maybe we may want to provide a tutorial or a short example? |
|
Let's wait on this, I'm still tweaking it and a tutorial would be nice too |
…te and 100x faster
|
Tutorial is added, ready for review. |
|
Alright, I think this is finally ready for review again @TParcollet @ycemsubakan @mravanelli . It now includes spectral features, and matches PRAAT and OpenSMILE. Later perhaps a recipe can be added for some open dataset. |
bcordel
left a comment
There was a problem hiding this comment.
Couple of comments for the jupyter notebook:
Title section:
This notebook goes through a simple voice analysis of a few speech samples. If you are new to speech processing, we recommend reading through this introduction before going through the notebook. First we download a public Parkinson's dataset and cut to just the sustained phonation.
Compute autocorrelation and related features (code box 1):
line 8: perhaps a comment/link about how to estimate best_lags
line 24: step_samples is hardcoded as 441
Could put the same comment from vocal_features.py into .ipynb for GNE
Maybe a "Here are some additional speech processing resources" box before the Speechbrain citation with (for ex):
https://tahull.github.io/blog/2020/08/acf-animated
https://github.com/chautruonglong/Fundamental-Frequency
https://www.fon.hum.uva.nl/praat/
https://www.audeering.com/opensmile/
No comments for the .pys, looks good to me !
|
Hi @mravanelli , @bcordel has completed his review and I was able to address the comments. I guess the last thing is your review, let me know if there's anything I can do to help. |
ycemsubakan
left a comment
There was a problem hiding this comment.
I think the additional explanations help a lot!
This PR introduces functions for helping with voice analysis, such as dysarthric speech detection.
The planned functions are as follows:
This provides a start, more may get added later. Tutorial included.