A high-performance Model Context Protocol server for concurrent audio-to-text transcription powered by GLM-ASR-2512.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install autoglmasr
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 autoglmasr 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 AutoGLM ASR MCP?
The AutoGLM ASR MCP server is a developer-centric tool designed to empower AI agents with advanced speech recognition capabilities. By utilizing the GLM-ASR-2512 engine, this skill provides a robust framework for transcribing audio files of any length. It is a cornerstone of the Openclaw Skills ecosystem for developers who need reliable, scalable, and context-aware transcription.
This skill excels at handling long-form audio by automatically splitting files into manageable chunks and processing them in parallel. It offers multiple context modes to balance speed and accuracy, making it suitable for everything from short voice commands to hour-long recordings.
AutoGLM ASR MCP Use Cases
- Transcribing long-form recordings like meetings, lectures, or podcasts for AI analysis.
- Building voice-enabled AI agents that require real-time or batch audio processing.
- Generating timestamped subtitles or segments for video and audio content.
- Automating data extraction from massive audio datasets using Openclaw Skills architecture.
How AutoGLM ASR MCP Works
- The AI agent triggers the transcribe_audio tool with a local file path.
- The server uses VAD (Voice Activity Detection) to identify silent periods and split the audio into logical chunks (max 25s).
- Silent segments are automatically skipped to reduce API costs and improve processing speed.
- Audio chunks are uploaded and processed concurrently using an asynchronous semaphore mechanism to respect rate limits.
- The system applies the selected context mode (Sliding, None, or Serial) to ensure linguistic coherence between segments.
- Results are merged into a single text output with accurate, offset-corrected timestamps for every segment.
AutoGLM ASR MCP Setup
To use this skill, ensure you have ffmpeg installed on your system. You can then configure the server within your MCP-compatible environment using your Zhipu API key.
# Install system dependency
brew install ffmpeg
# Run via npx
npx autoglm-asr-mcp
Add the following to your MCP configuration file:
{
"mcpServers": {
"autoglm-asr": {
"command": "npx",
"args": ["-y", "autoglm-asr-mcp"],
"env": {
"AUTOGLM_ASR_API_KEY": "your-api-key"
}
}
}
}
AutoGLM ASR MCP Data Schema & Taxonomy
The skill exposes two primary tools: transcribe_audio and get_audio_info. It handles various metadata and operational statistics to give developers full visibility into the transcription process.
Tool Parameters: transcribe_audio
| Parameter | Type | Required | Description |
|---|---|---|---|
audio_path |
string | Yes | Absolute path to the audio file |
context_mode |
string | No | sliding (default), none, or full_serial |
max_concurrency |
int | No | Number of parallel requests (1-20) |
Output Structure
- Full Text: The complete reconstructed transcript.
- Segments: List of objects containing
start,end, andtextfor each phrase. - Stats: Execution metrics including chunk count, time elapsed, and skipped silence segments.
AutoGLM ASR MCP Advanced Features
- Context Passing: Use sliding mode to maintain state between audio chunks for better accuracy in continuous speech.
- High Concurrency: Process long files up to 20x faster by utilizing parallel API requests through Openclaw Skills integration.
- Cost Optimization: Built-in VAD (Voice Activity Detection) ensures you only pay for segments containing actual speech.
- Wide Format Support: Native handling of mp3, wav, m4a, flac, ogg, and webm files.
- Automatic Stereo-to-Mono Conversion: Resolves common ASR API errors by pre-processing audio to the required single-channel format.
SKILL.md
Loading
Related Openclaw Skills
Xiaohongshu Content Creator
An AI-driven content architect designed to generate high-engagement titles, structured notes, and optimized hashtag strategies for Xiaohongshu creators.
Neurology Medical Record Structuring Assistant
An AI-driven assistant that converts raw outpatient clinical text into standardized, structured neurological medical records.
Medical Record Structuring Assistant
A clinical documentation tool that transforms raw outpatient and inpatient notes into professional, structured medical summaries.
Medical Document Processor
An AI-driven assistant tailored for healthcare professionals to automate medical history summarization and clinical report interpretation.
Find Skills
A command-line utility to discover, install, and manage new capabilities for your AI agent through the Openclaw Skills marketplace.
LiteLLM
LiteLLM provides a unified API interface to call over 100 LLM providers seamlessly within your AI agent workflows.
Featured*
