Live intelligence

Latest AI updates

2 updates
Models

ZONOS3-Mini: Ultra-Compact Zero-Shot Voice Cloning TTS with Open Weights

ZONOS3-Mini is an ultra-compact, real-time zero-shot voice-cloning TTS model designed to run on laptops or smartphones. It uses a latent mixture-of-experts backbone with 886M parameters (100M active per token) and a new audio codec, Zyphra Audio Codec 1, representing 44.1 kHz audio at a 15 Hz frame rate. On the ZTTS2-Eval benchmark, it achieves a 6.71% macro-average WER across 55 language sets, competitive with larger and slower models. The authors release model weights, inference code, and training code.

ZONOS3-MiniZyphra Audio Codec 1ZTTS2-Eval
Models Editor selected

Zyphra Releases ZONOS3-Mini: Open-Source On-Device TTS With 100M Active Parameters

Zyphra has released ZONOS3-Mini, an Apache 2.0 text-to-speech model with 800M total parameters and about 100M active parameters in a sparse Mixture-of-Experts design. It supports zero-shot voice cloning, word-level delivery control, and deployment via WebGPU in browsers, SGLang-Omni on NVIDIA GPUs, MLX on Apple Silicon, and ggml across CPU, CUDA, Metal, and Vulkan. Zyphra reports 80 ms median time to first audio on one H100 with eight concurrent requests, while browser latency on a Pixel 9 is under 570 ms. The model tops Zyphra's self-run ZTTS2-Eval Capability Index at 66.2 in speaker-embedding mode, but reference-clip cloning scores lower and all benchmarks are self-reported.

Why it matters: It gives AI developers a concrete, deployable open-source TTS release with architecture, runtime support, and benchmark caveats relevant to on-device voice applications.

ZyphraZONOS3-Minitext-to-speech