qwen-tts

What it is and what it does

Qwen-TTS is a Python wrapper for Alibaba's Qwen3-TTS speech generation models. It lets you convert text to speech in 10 languages (Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish, Italian) with fine-grained control over voice characteristics. The package supports three main workflows: custom voice generation from text descriptions, voice cloning from a 3-second audio sample, and instruction-based voice design where you describe desired acoustic properties (timbre, emotion, prosody) and the model adapts output accordingly.

The underlying models use a discrete multi-codebook language model architecture for end-to-end speech generation, avoiding cascading errors from traditional pipeline approaches. Both streaming (output starts after a single character) and non-streaming modes are supported. Runtime dependencies include transformers, torchaudio, librosa, and onnxruntime for inference, plus gradio for the optional web UI. Model weights download automatically from Hugging Face or ModelScope on first use.

Use it for:

• Generate natural-sounding voiceovers for videos or podcasts in multiple languages with custom voice profiles.
• Clone a speaker's voice from a short audio clip and use it to synthesize new speech in the same voice.
• Build interactive voice assistants that respond with emotionally appropriate speech based on text instructions.
• Create accessible audio versions of documents by converting text to speech with fine-tuned prosody and tone.
• Prototype multilingual speech applications without training custom models from scratch.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Qwen-TTS generates speech from text using Qwen3-TTS models, supporting voice cloning, voice design, and instruction-based voice control across 10 languages with streaming and non-streaming output.

Yes, if you need multilingual text-to-speech with voice cloning and instruction-based control. The low install friction and permissive license are favorable. However, the package is very recent (first release 2026-01-22) with aging maintenance status, so expect potential API changes and monitor for updates. Test thoroughly before production use, especially for latency-sensitive applications.

Install

qwen-tts on PyPI

pip install qwen-tts

uv add qwen-tts

poetry add qwen-tts

Installing qwen-tts

Before you install

Low friction: pure Python wheel with no compiled dependencies. Aging status (189 days since release) suggests early-stage maintenance; monitor for updates as the package matures.

License in practice

Apache-2.0 permissive license allows commercial and private use with minimal restrictions; attribution required but no copyleft obligations.

Quickstart

pip install qwen-tts

from qwen_tts import TTS

tts = TTS(model="Qwen3-TTS-12Hz-1.7B-CustomVoice")
audio = tts.generate("Hello world")

Requires Python 3.9+; model weights auto-download on first use (can be pre-downloaded via modelscope or huggingface-cli if network-constrained).

Verify before relying

• Actual latency and quality characteristics in production deployments (description claims 97ms end-to-end but real-world variance is unknown).
• Memory and GPU requirements for each model variant (0.6B and 1.7B sizes mentioned but resource profiles not detailed).
• Stability and API surface maturity given first_release is 2026-01-22 (very recent).

Package facts

Evidence: qwen_tts-0.1.1-py3-none-any.whl

Tags

More Artificial Intelligence packages