introlab / odas_ros

ODAS ROS package
GNU General Public License v3.0
39 stars 23 forks source link

Matching sound separation sources with sound tracked sources #18

Open aandriella opened 2 years ago

aandriella commented 2 years ago

Hello, I'm working on integrating odas_ros into our ros4hri pipeline.

As you can see, to each human, we would like to assign a and the audio source to improve the speech recognition. To my understanding, the current version of the sound source separation seems to output only an AudioFrame which contains up to 4 different sources but it does not match them to those that are tracked. Is that correct? If so, are you planning to develop it in the near future?

Thank you.