A new, compact English text-to-speech model named tiny-voca has been released, featuring two distinct voices and a waveform generator with 4.5 million parameters. This model operates significantly faster than real-time, approximately 70 times faster on four CPU threads, and was trained entirely on synthetic speech data generated by Qwen3-TTS. The model is available for use with PyTorch and ONNX runtimes, including onnxruntime-web for browser-based applications. AI
IMPACT Offers a fast and efficient text-to-speech solution for developers and researchers.
RANK_REASON Release of a new, small-scale AI model with technical details. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Trending Models →
- facebook/wav2vec2-lv-60-espeak-cv-ft
- n4ze3m/tiny-voca
- ONNX
- onnxruntime
- onnxruntime-web
- PyTorch
- Qwen3-TTS
- tiny-voca
- Whisper
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →