A high-performance speech-to-text skill powered by the SiliconFlow SenseVoice API for seamless audio transcription.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install speech-recognition
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 speech-recognition 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 Speech Recognition?
The Speech Recognition skill provides Openclaw agents with the ability to transcribe various audio formats into highly accurate text. By leveraging the SiliconFlow SenseVoice API, it excels at processing multilingual audio, particularly Chinese and English speech. This skill is a foundational tool within the Openclaw Skills ecosystem for bridging the gap between voice-based input and text-based AI processing.
Whether handling direct voice messages or processing uploaded audio files, this skill ensures that agents can understand and act upon verbal information. It handles the complexities of audio sampling and format normalization, making it an essential component for any voice-enabled workflow.
Speech Recognition Use Cases
- Transcribing voice messages from communication platforms like Telegram or Signal.
- Converting recorded voice memos and meeting audio into searchable text documents.
- Extracting dialogue from media files to be used as context for other Openclaw Skills.
- Automating the transcription of user-uploaded .mp3 or .wav files for data analysis.
How Speech Recognition Works
- The skill is triggered when a user sends an audio file or explicitly requests a transcription.
- It identifies the audio format; if the file is in a non-standard format like OGG, it uses FFmpeg to convert it to an optimized MP3 or WAV file.
- The processed audio is sent to the SiliconFlow SenseVoiceSmall API endpoint using the configured API key.
- The API performs speech-to-text inference and returns a JSON response containing the transcription.
- The skill extracts the text content and presents it to the user or passes it to the next agent in the workflow.
Speech Recognition Setup
To use this skill, ensure you have FFmpeg installed for audio processing and a valid SiliconFlow API key. Add your credentials to the configuration file at ~/.openclaw/openclaw.json:
{
"providers": {
"siliconflow": {
"apiKey": "sk-your-api-key-here"
}
}
}
You can also set the SILICONFLOW_API_KEY environment variable for quick testing.
Speech Recognition Data Schema & Taxonomy
| Component | Details |
|---|---|
| Supported Formats | .mp3, .ogg, .wav, .m4a, .flac |
| AI Model | FunAudioLLM/SenseVoiceSmall |
| API Endpoint | https://api.siliconflow.cn/v1/audio/transcriptions |
| Audio Requirements | Recommended < 10MB file size and < 5 minutes duration |
| Processing Specs | 16kHz sampling rate, Mono channel |
Speech Recognition Advanced Features
- Automatic format conversion using FFmpeg for seamless Telegram/Signal voice message handling.
- Native integration with other Openclaw Skills like douyin-video for audio extraction and transcription.
- Intelligent error handling for network timeouts and payload size limits.
- Support for specialized audio parameters including custom sampling rates and mono-channel downmixing.
SKILL.md
Loading
Related Openclaw Skills
Self-Reflection Skill
A sophisticated self-audit and evolution system for AI agents based on role theory, systems thinking, and cognitive bias mitigation.
Discord Context Sync
A utility to synchronize, cache, and link Discord Forum thread discussions into localized memory for AI agent context.
Stock Evaluator v3
An institutional-grade stock analysis tool for AI agents that delivers comprehensive fundamental research, technical assessments, and clear investment recommendations.
unfuck-my-git-state
A staged, low-risk recovery framework for diagnosing and fixing corrupted Git repository metadata and inconsistent worktree states.
sugarclawdy
A comprehensive tool for registering AI agents on the SugarClawdy platform to enable verification and reward systems.
ClawNet
ClawNet provides a P2P discovery and communication protocol for OpenClaw agents to find and interact with each other securely over the internet.
Featured*
