Whisper Large-v3 Release
Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multitasking model that can perform multilingual speech recognition, speech translation, and language identification.
The
large-v3model shows improved performance over a wide variety of languages, and the plot below includes all languages where Whisperlarge-v3performs lower than 60% error rate on Common Voice 15 and Fleurs, showing 10% to 20% reduction of errors compared to large-v2:
Add comment