A comprehensive framework for configuring, debugging, and optimizing local OpenAI-compatible text-to-speech servers.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install local-tts-workflow
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install local-tts-workflow using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Local TTS Workflow?
The Local TTS Workflow is a specialized skill designed to manage the complexities of local speech synthesis within Openclaw Skills environments. It provides developers with a structured approach to interface with self-hosted TTS servers, such as vLLM Omni or MLX-Audio, ensuring that audio generation is reliable, low-latency, and natural-sounding. By focusing on both the infrastructure layer and content preparation, this skill bridges the gap between raw AI inference and high-quality voice output.
Central to this skill is the ability to validate the entire speech pipeline, from the initial OpenClaw configuration to the final streaming audio transport. It emphasizes the importance of text normalization, ensuring that digits and symbols are converted into spoken words before synthesis to prevent model glitches. Whether you are implementing voice cloning, custom speaker designs, or real-time streaming, this workflow ensures your local AI agents communicate clearly and effectively.
Local TTS Workflow Use Cases
- Debugging connectivity and request shapes for local v1/audio/speech endpoints.
- Normalizing dates, times, and numeric counts into spoken words for natural synthesis.
- Validating voice cloning workflows that require synchronized reference audio and text.
- Testing real-time streaming audio performance using CLI-based playback tools.
- Configuring multi-mode synthesis including Base Speaker, Voice Clone, and Voice Design.
How Local TTS Workflow Works
- Inspect the local OpenClaw configuration file to identify the active TTS provider, model, and base URL.
- Execute a connectivity check against the server's health and model inventory endpoints to ensure availability.
- Normalize the input text by converting all Arabic numerals and symbols into their spoken word equivalents.
- Determine the required synthesis mode (Base, Clone, Custom, or Design) and construct the corresponding JSON payload.
- Perform an isolated curl test to verify that the server returns valid audio independent of the agent integration.
- Compare the local OpenAPI schema against the request shape to identify parameter mismatches or missing fields.
Local TTS Workflow Setup
To get started, verify your local OpenClaw configuration and ensure your TTS server is reachable. Use the following commands to check your environment:
# Check the active OpenClaw TTS configuration
cat ~/.openclaw/openclaw.json | grep tts
# Verify server health and available models
curl http://127.0.0.1:8000/health
curl http://127.0.0.1:8000/v1/models
If using a local MLX server, ensure the TTS task is explicitly enabled:
MLX_AUDIO_SERVER_TASKS=tts uv run python server.py
Local TTS Workflow Data Schema & Taxonomy
The workflow relies on a structured hierarchy of configuration and API interactions to ensure consistency across Openclaw Skills deployments:
| Component | Source/Endpoint | Purpose |
|---|---|---|
| Config File | ~/.openclaw/openclaw.json |
Primary source of truth for provider and model settings. |
| TTS API | /v1/audio/speech |
Main endpoint for both streaming and non-streaming synthesis. |
| Consent API | /v1/audio/voice_consents |
Manages reference audio and text for voice cloning modes. |
| Local Models | ./models/ |
Fallback directory for local-path-backed model deployments. |
| Documentation | references/tts-api.md |
Local schema for exact field names and streaming semantics. |
Local TTS Workflow Advanced Features
- Real-time incremental streaming support with adjustable streaming intervals for low-latency feedback.
- Advanced text normalization logic to handle mixed-language numeric snippets (e.g., Chinese and English).
- Multi-mode request handling for VoiceDesign (instruction-driven) and CustomVoice (prebuilt speakers).
- Consent-based cloning workflows requiring strict mapping between reference audio and reference text.
- Support for single-reply overrides using the [[tts:...]] tag system within OpenClaw interactions.
SKILL.md
Loading
Related Openclaw Skills
OpenClaw Deploy Tracker
A specialized tracking system for logging, querying, and managing service deployments within the OpenClaw ecosystem.
10jqka-concept
A specialized financial analysis skill for scraping and parsing stock concept and sector information from Tonghuashun (10jqka) F10 pages.
v0-cli
A specialized command-line interface for AI agents to programmatically generate and refine modern web interfaces using v0.dev.
Alibaba Cloud KMS Secret Management
A comprehensive AI skill for managing the lifecycle of sensitive credentials and secrets within Alibaba Cloud Key Management Service.
Yintai Tasks Runner
An automated task management tool for the Yintai platform that handles task grabbing, status tracking, and standardized delivery generation.
USMLE Case Generator
A clinical scenario engine designed to produce high-fidelity USMLE Step 1 and Step 2 CK style vignettes for medical students and educators.
Featured*
