Audio8 TTS Preview 0.1B Releases Compact Multilingual Speech Model

Summary

Audio8 TTS Preview 0.1B is a compact multilingual text-to-speech model for speech generation and zero-shot voice cloning. The release includes an approximately 170M-parameter generative model, neural audio codec, tokenizer, processor, Transformers remote code, and an ONNX INT8 variant. Its architecture combines a Falcon-H1 hybrid backbone using Mamba 2 and attention with a faster branch that predicts codec codebooks. The supplied article does not report benchmarks, language coverage, licensing, or latency measurements.