Audio8 releases compact 0.6B TTS model with zero-shot voice cloning

Audio8 has released a new text-to-speech model, Audio8 TTS Preview 0.6B, which is notable for its compact size and SOTA-class performance. Despite its 0.6 billion parameters, the model achieves competitive results on benchmarks, including the best English Word Error Rate (WER) on the Seed-TTS benchmark. It supports multilingual speech generation and zero-shot voice cloning, with plans for broader language support in future releases. AI

IMPACT This compact TTS model could enable more efficient deployment of high-quality speech generation in resource-constrained environments.

RANK_REASON Model release from a known AI lab (Audio8) with specific performance claims. [lever_c_demoted from frontier_release: ic=1 ai=1.0]

Read on Hugging Face Trending Models →

• Audio8/Audio8-TTS-Preview-0.6b
• Audio8 TTS Preview 0.6B
• AutoArk-AI
• DualAR
• Fish Audio S2 Pro
• Hugging Face
• PyTorch
• safetensors
• soundfile
• torchaudio
• transformers

• model release
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →