SpeechLink
We finetuned the Whisper-Small model for speeches with dysarthria based on the public TORGO dataset(https://www.cs.toronto.edu/~complingweb/data/TORGO/torgo.html) by UofT.
This model achieved an 81% improvement from a WER of 1.73 to 0.33.
- Downloads last month
- -
Model tree for feiiyw/Speechlink-v1
Base model
openai/whisper-small