Skip to content
TakoAPI
Back to skills

local-llama-tts

by wuxxin

AI & LLMs
430Downloads reported by clawskills.sh
0Views
0 likesSign in to like this skill
—Not yet rated
Your rating

Brief

Local text-to-speech using llama-tts (llama.cpp) and OuteTTS-1.0-0.6B model.

Description

What This Skill Does

Runs text-to-speech synthesis entirely on local hardware using llama.cpp and the OuteTTS-1.0-0.6B model. Produces WAV audio files from text input with optional speaker reference for voice cloning. No network calls or external APIs involved.

Runs fully offline with GPU acceleration, so there are no API costs, rate limits, or data privacy concerns compared to cloud TTS services.

When to Use It

  • Converting notes or documents to audio for offline listening
  • Generating voiceovers for local video projects
  • Testing TTS output with different speaker reference files
  • Batch-producing spoken audio from script text
  • Running speech synthesis on air-gapped machines

Example Workflow

INPUT

User asks: convert this paragraph to speech and save it as narration.wav

AGENT
  1. 1

    Receives the text and target filename

  2. 2

    Calls scripts/tts-local.sh with the text and -o narration.wav

  3. 3

    llama-tts loads OuteTTS-1.0-0.6B-Q4_K_M.gguf and WavTokenizer models from local paths

  4. 4

    Synthesizes audio using configured sampling parameters (temp 0.4, top-p 0.9, rep penalty 1.1)

  5. 5

    Writes the result to narration.wav

OUTPUT

narration.wav file containing synthesized speech

Requirements

  • llama-tts binary installed and available in PATH
  • OuteTTS-1.0-0.6B-Q4_K_M.gguf model file downloaded from HuggingFace (OuteAI/OuteTTS-1.0-0.6B-GGUF)
  • WavTokenizer GGUF vocoder model downloaded from HuggingFace (ggml-org/WavTokenizer)

Installation

clawhub install wuxxin/local-llama-tts

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

Links