Can I run this on a Mac M2?

Yes, this skill is specifically optimized for Apple Silicon (Mac M2+) using the MLX framework for high-performance inference.

Why is my generated audio too long or taking forever to process?

Qwen3-TTS defaults to 4096 tokens (approx. 327s). This skill recommends setting max_tokens to 150 for standard 12-second clips to ensure efficient generation.

What is RTF in the context of this skill?

Real-Time Factor (RTF) measures processing speed. An RTF < 1 means the audio generates faster than its playback duration, making it suitable for live applications.

How do I adjust the speech speed in Qwen3-TTS?

Qwen3-TTS VoiceDesign doesn't support speed parameters directly in the prompt; it is recommended to adjust the playback rate during audio rendering.

Qwen3-TTS MLX Optimizer

Name: Qwen3-TTS MLX Optimizer
Author: kazuph

bykazuph

•

Data Science & ML

Optimizes Qwen3-TTS performance on Mac MLX with specialized parameter tuning, voice design presets, and integration guides.

This skill provides comprehensive guidance and implementation patterns for running Qwen3-TTS in VoiceDesign mode on Mac M2+ hardware using the MLX framework. It solves common issues like excessive token generation and performance bottlenecks by providing precise RTF (Real-Time Factor) benchmarks, dynamic token calculation formulas, and ready-to-use Python snippets for high-quality voice synthesis. It is particularly useful for developers building AI-powered narrators, interactive Web UIs, or automated video production workflows with Remotion.

Key Features

01Real-Time Factor (RTF) benchmarks for local vs. cloud environments

0215 GitHub stars

03Integration blueprints for FastAPI Web UIs and Remotion video workflows

04Dynamic max_tokens calculation to prevent audio clipping or excessive generation

05Optimized parameter presets for Mac M2+ MLX hardware performance

06Ready-to-use voice design presets for various personas and moods

Use Cases

01Building interactive web applications with low-latency text-to-speech

02Automated video narration and content creation for social media

03Local AI voice generation for Mac-based application developers

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add kazuph/dotfiles qwen-tts

For use in Claude.ai and ChatGPT

Download Skill