About this skill

Problem

When engineers need to generate speech from scripts, CI jobs, or toolchains instead of using a web console, the practical issue is how to call MiMo V2-TTS consistently and save or stream the audio output. This skill packages that workflow into CLI arguments, reducing repeated API request and format-handling code.

How it works

I/O: pass the prompt text with --text, specify the file path with --output; default output is wav, while streaming uses pcm16.
Voice and style: select mimo_default, default_zh, or default_en with --voice; the synthesis text is placed in an assistant-role message, style tags must start the text, and the 唱歌 tag must be used only where singing mode is required.
Runtime requirements: set MIMO_API_KEY; keep each request under about 2000 characters and split longer text to avoid timeouts.

Fit and limits

It is useful for scripted Chinese, English, or default-voice audio generation and for feeding TTS output into downstream pipelines. Complex multi-speaker orchestration, ultra-low-latency realtime interaction, or undocumented style tags need separate validation.

Screenshots

Use Cases

Convert podcast scripts into Chinese female-voice WAV files for import into editing software.
Generate English female-voice WAV output for automated news broadcast scripts and save it to a target directory.
Split long text into TTS requests in a local toolchain and produce multiple WAV segments for post-processing.
Validate realtime speech pipelines using streaming mode and pcm16 audio output.

Best For

Backend engineers who need to convert script or CI text into WAV files
Podcast producers who need Chinese and English female-voice demo audio
Frontend or embedded engineers building voice assistants and needing pcm16 streaming output
Content operations staff who need scripted voiceover assets with selected voices and styles