From Common Voice to Common Sounds

Just had an idea the other day while having a hike in the beautiful Swiss countryside. Iˋm living close to a small airfield just outside of Zurich. While walking in the forest, I heard a sound that I couldnt recognise. Was it a plane? Was it a car? I wasnˋt sure. Hence, I thought, what if systems could learn sounds? This might trigger new applications besides speech recognition (e.g. just thinking of drones, self-driving cars or any other artificial agent that needs to make sense of its environment).

The classification of sounds could e.g. also be made by the observer / listener and double-checked for consistency just like in Common Voice.

Note: Haven’t checked what’s already out there.

Actually there a dataset like that it’s called Urban sounds classification, although it has only about 8k samples.

1 Like

Thanks for checking.

I’m moving this to #deep-speech since it seems more like a ML question, out of the scope of common voice.

Thanks.