Whistle: Speech to Text in 16.9 MB
AI summary
Editorial policyWhistle, a 16.9MB open-source speech recognition model, runs on the same CPU engine as Needle and enables tool calling.
Key points
- •Whistle is an open speech recognition model that runs on the same CPU engine as Needle.
- •It supports transcription in seven languages and reaches the first token generation in 11 milliseconds.
- •By loading it in parallel with Needle, clips can be converted into tool calls within a single binary.