Brief

HiggsAudio-Studio has been released as a portable, offline text-to-speech solution that supports over 100 languages and features zero-shot voice cloning. The studio is built on Higgs Audio v3 TTS and includes AI-powered text direction, along with podcast and audiobook modes. While a version 2 training script exists, users are reporting issues with a manual LoRA training script for version 3. AI

Boson AI has released Higgs Audio v3 TTS, a text-to-speech model designed for conversational voice chat. The model supports over 100 languages, offering zero-shot voice cloning and fine-grained control over emotion, style, and prosody. It utilizes an autoregressive decoder with interleaved text and audio tokens, encoding audio into codebooks for processing. While released for research, commercial use requires a separate license, with strict prohibitions against unlawful applications. AI