I want to detect if someone is speaking (human voice detect)

Question/Issue:
I inser in my project only speech datasets (speech label) but if I use Impulse = classifier it asks me for at least two labels. Is it mandatory to include a second class? Is there a better method to detect Human voice?

Project ID:
390809

Context/Use case:
Sound event detection with multi impulse
One impulse with MFE for sed
One impulse with MFCC for Human Voice Detect

Yes, just add another class with different background sounds, without having speech in them. Another way might be to use anomaly detection, but I’m unsure if it’s available for sound data.

1 Like

ok Thomas thanks a lot

1 Like