Skip to content
TakoAPI
Back to skills

qwenspeak

by psyb0t

Media & Streaming
585Downloads reported by clawskills.sh
0Views
0 likesSign in to like this skill
—Not yet rated
Your rating

Brief

Text-to-speech generation via Qwen3-TTS over SSH.

Description

What This Skill Does

YAML-driven text-to-speech over SSH using Qwen3-TTS models. Supports three modes: preset voices, voice design from natural language descriptions, and voice cloning from reference audio. Jobs are submitted as YAML configs, run sequentially, and output WAV files.

Runs on self-hosted infrastructure via SSH, keeping audio data private and avoiding per-request cloud TTS costs.

When to Use It

  • Generating narration for tutorial or explainer videos
  • Cloning a specific person's voice for consistent audio content
  • Batch-producing dialogue lines for a game or animation
  • Designing a custom voice character from a text description
  • Producing multilingual voiceovers from a single YAML workflow

Example Workflow

INPUT

User asks: clone my voice from a reference recording and read out a three-line script

AGENT
  1. 1

    Upload the reference audio file using 'put refs/me.wav'

  2. 2

    Write a YAML job with voice-clone mode, setting ref_audio to refs/me.wav and ref_text to the transcript

  3. 3

    Submit the job via the 'tts' command and capture the returned job UUID

  4. 4

    Poll job status with 'tts get-job <uuid>' until status reaches completed

  5. 5

    Download each output WAV file using 'get <output_path>'

OUTPUT

WAV files with the cloned voice reading each line of the script

Requirements

  • QWENSPEAK_HOST environment variable
  • QWENSPEAK_PORT environment variable
  • Running qwenspeak instance (self-hosted via https://github.com/psyb0t/docker-qwenspeak)

Installation

clawhub install psyb0t/qwenspeak

Run the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.

Links