Test feature extraction tools for audio files

[yibeichan]

Objectives : Extract interesting timepoints from audio (big noise, call out, etc,...) to map them to the videos and further quantify child's joint attention or reactivity
Technologies : Checkout senselab and litterature
Blocking points : None

[lucie271]

I imported senselab library.
Would need to dig deeper in :

• getting interesting timepoints (name calling,..) from conversations extracted (working but can be way imroved atm)
• identifying louder noises/conversation for initation of Joint Attention moments

[lucie271]

Need to improve:

• name calling identification accuracy (misses a lot of name calinngs)
• improve the according timepoint extraction

No need to identify louder noises/conversations for the moment (not in annotations)