qwen-tts_skill

Overview

This skill provides local text-to-speech using the Qwen3-TTS-12Hz-1.7B-CustomVoice model. It runs entirely offline after the initial model download and supports 10 languages, 9 premium speaker voices, and instruction-based voice control for emotion, tone, and style. Use it as a privacy-friendly alternative to cloud TTS services when you need reliable, on-device audio generation.

How this skill works

The skill installs a small runtime and downloads the 1.7GB TTS model on first use. A command-line script accepts text, language, speaker, and an instruction string to shape emotion and style, then synthesizes WAV audio locally (16 kHz). It auto-detects available hardware, preferring GPU if available and falling back to CPU when memory is limited.

When to use it

• Generate offline audio files from text for apps or demos
• Create voice messages, narration, or audio prompts without cloud services
• Control emotional tone, style, and pacing via instruction strings
• Produce speech in supported languages including Italian for local projects
• Integrate with automation pipelines that capture output file paths

Best practices

• Use a speaker matched to the text language for best quality (native-language speakers recommended)
• Provide concise instruction strings (emotion, tone, style) to shape delivery
• Pre-run setup on target machines to avoid model download delays during production
• Prefer GPU for faster synthesis; expect longer generation times on CPU
• List speakers with --list-speakers when choosing a voice

Example use cases

• Generate Italian narration: synthesize onboarding messages or tutorials in Italian using the Italian-capable speakers
• Create dynamic IVR prompts: produce multiple versions with varying emotion and tone using instruction-based control
• Localizable game dialogue: synthesize character lines offline with different premium voices
• Privacy-sensitive voice output: produce customer-facing audio without sending text to cloud services
• CI/CD audio asset generation: script TTS to produce updated audio files and return path for automation