Cactus Compute has released Whistle, a new speech-to-text model designed for efficient on-device operation across various platforms like mobiles, wearables, and microcontrollers. The model is a compact 16.9 MB file and runs on the same CPU engine as their previous Needle model, requiring no external dependencies or GPUs. Whistle offers transcription in seven languages, word-level timestamps with probability, and speech embedding for matching and retrieval, with an option for keyword biasing to improve accuracy for specific terms. AI
IMPACT Enables more efficient and capable on-device speech processing for a wider range of applications.
RANK_REASON Release of a new speech-to-text model with detailed technical specifications and benchmark comparisons. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Trending Models →
- Amazon Machine Image
- Apple M4 Pro
- Cactus-Compute/needle3
- Cactus-Compute/whistle
- Earnings-22
- Moonshine
- rock
- SPGISpeech
- TED-LIUM corpus
- Whisper
- whistle
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →