- Inflect-v2 introduces two exceptionally small English TTS models with just 3.9M and 9.3M parameters.
- Both models generate speech multiple times faster than real-time on CPU.
- Despite their small size, Inflect-v2 delivers quality comparable to larger lightweight TTS systems like KittenTTS, Piper, and Supertonic-3.
- The models support CPU, CUDA, PyTorch, and ONNX, and are released under Apache 2.0.