AI BriefingFlare-TTS 28M released
·2026.05.02 19:52
Key point
Flare-TTS 28M, trained for 24 hours on a single A6000, has been released.
Details
Flare-TTS 28M has been released.
- Trained for about 24 hours on a single A6000 GPU.
- Trained from scratch for 300 epochs on the full LJSpeech dataset.
- A 28M-scale TTS model that generates English speech.
- Released as free and open-source on Hugging Face.
- Still leans somewhat toward a robotic tone.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Inflect-Micro-v2: A Complete Voice Model Implemented with 9.36M Parameters
Supertonic 3 Launches as Ultra-Lightweight On-Device TTS, Supporting 31 Languages and Emotion TagsVoxCPM2 - Open-source multilingual speech synthesis with voice cloning that sounds like real speech