Local zero-shot voice cloning with Qwen3-TTS-12Hz-1.7B-Base
Melvin Vivas · X post · 2026-10-23 · Open on X
Topics: AI Dev Tools & Productivity, LLM Fundamentals · Level: intermediate
Summary
The creator recommends Qwen3-TTS-12Hz-1.7B-Base for voice cloning. It does zero-shot cloning from a 10-second reference clip, needs no fine-tuning, and runs locally.
Key points
- Model: Qwen/Qwen3-TTS-12Hz-1.7B-Base on Hugging Face (1.7B parameters).
- Zero-shot voice cloning from about a 10-second reference audio clip.
- No fine-tuning needed to clone a new voice.
- Can be run locally, with no cloud API required.
Resources mentioned
- Qwen3-TTS-12Hz-1.7B-Base · tool · huggingface.co · free
Open-weight Qwen text-to-speech model on Hugging Face that does zero-shot voice cloning and runs locally.
Try this
- Try Qwen3-TTS-12Hz-1.7B-Base locally to clone a voice from a 10-second reference clip.
- Build a local voice-cloning text-to-speech app using Qwen3-TTS with a short reference recording.