Use Uttera to transcribe, analyze, translate, and generate speech and audio through one Openclaw skill.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install uttera
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 uttera 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 Uttera Audio & Voice?
Uttera Audio & Voice connects an OpenClaw coding agent to Uttera’s voice AI API for speech-to-text, meeting summaries, translation, speaker diarization, pronunciation scoring, text-to-speech, voice cloning, and generated sound effects, music, and sound scenes. It also creates signed PDF reports and verifies their authenticity.
The bundled Python script sends the supplied audio directly to Uttera and returns the result; the agent does not decode or analyze audio itself. This makes the skill useful for audio workflows that need a dedicated API, while keeping billing, account requirements, data handling, and voice-cloning consent in view. Discover practical Openclaw Skills for audio and voice automation with Uttera.
Uttera Audio & Voice Use Cases
- Transcribe an attached recording into plain text.
- Summarize a call or meeting with key points, action items, speaker information, and tone.
- Create a signed PDF report with an evidence graph for a recording.
- Translate speech into another language as text or spoken audio.
- Identify who spoke when using timestamped diarization.
- Score a short recording against reference text to assess pronunciation.
- Read supplied text aloud, or generate speech from an explicitly consented voice sample.
- Create sound effects, music, or a mixed sound scene from a text description.
- Check available voices, inspect the cost of the last call, or verify a signed report.
How Uttera Audio & Voice Works
- The agent identifies the requested action and collects its required inputs, such as an audio file, target language, reference text, or sound description.
- It runs the bundled
scripts/uttera.pyscript with the selected action and arguments. Audio-input actions pass the real file path to the script. - The script sends the file and API key over HTTPS to
api.uttera.ai. Audio is processed by Uttera; the agent must not attempt to decode, transcribe, summarize, or describe it independently. - Uttera returns text or structured JSON, or the script saves an audio or PDF output. The agent shows the script output and makes any saved file available to the user.
- The action is billed according to the Uttera account plan and credits. If an action is not included in the plan, the API returns a 403 plan message; the user may need to upgrade at
https://app.uttera.ai/billing.
Uttera Audio & Voice Setup
- Install or place the skill so its
scripts/uttera.pyfile is available, and ensurepython3is installed. - Configure an Uttera API key. The script checks
scripts/key.txt, theUTTERA_API_KEYenvironment variable, or~/.uttera/api_key. Calls use credits from the associated Uttera account. - Ensure the runtime can reach
api.uttera.ai. For Claude Team or Enterprise, an administrator may need to add that domain under Organization settings → Capabilities → Allow network egress → Domain allowlist. Individual Claude Free/Pro plans do not require this allowlist change. - Run the script from the skill directory, passing the action and required arguments:
python3 scripts/uttera.py transcribe "/path/to/recording.m4a"
python3 scripts/uttera.py summarize "/path/to/meeting.mp3"
python3 scripts/uttera.py translate "/path/to/audio.ogg" en
python3 scripts/uttera.py speak "Hello world" dora en
python3 scripts/uttera.py cost
Audio files must fit the sandbox upload limit of approximately 30 MB; larger files should be processed through the Uttera web app or API. Voice cloning and translating in the original speaker’s voice require the speaker’s explicit consent.
Uttera Audio & Voice Data Schema & Taxonomy
The skill accepts action-specific inputs and returns text, JSON, or a saved file. It does not describe a separate local database or persistent metadata store.
| Action | Input | Result |
|---|---|---|
transcribe |
Audio path | Plain text transcript |
summarize |
Audio path | Transcript and structured summary |
report |
Audio path | Saved signed PDF and printed summary; report includes an evidence graph when supported by the plan |
translate |
Audio path, target language code | Translated text |
translate-voice |
Audio path, target language code, optional voice or original |
Saved translated audio |
diarize |
Audio path | JSON speaker segments with timestamps |
pronunciation |
Audio path, reference text, optional language | JSON pronunciation scores, expected and heard IPA, and grouped errors |
clone |
Voice-sample path, text to speak, optional language | Saved generated audio |
speak |
Text, optional voice and language | Saved generated audio |
sfx, music, scene |
Description and action-specific duration or steps | Saved generated audio |
voices, cost, verify |
No argument, or a report JSON path for verify |
JSON voice list, last-call cost, or verification verdict |
Language arguments use codes such as en, es, fr, de, zh, or no. report saves a PDF; translate-voice, clone, speak, sfx, music, and scene save audio. verify checks a report’s .json file using Ed25519 verification. The script prints saved paths so the agent can surface the resulting files.
Uttera Audio & Voice Advanced Features
- Supports 14 actions spanning transcription, summarization, translation, diarization, pronunciation assessment, speech generation, audio generation, billing inspection, and report verification.
- Produces signed PDF reports with an evidence graph and provides an Ed25519 verification action for report integrity.
- Supports translated speech in a catalogue voice or, with consent, the original speaker’s cloned voice.
- Generates sound effects up to 30 seconds, music up to 380 seconds, and timed sound scenes. Music cost depends on duration multiplied by generation steps.
- Provides JSON outputs for diarization, pronunciation analysis, available voices, cost details, and verification results.
- Sends the audio and API key only to
api.uttera.aiover HTTPS; the API base override is restricted to*.uttera.ai. - Generated audio includes an inaudible watermark required by the EU AI Act and cannot have it disabled.
- Voice cloning uses biometric personal data. Only clone the user’s own voice or a voice whose speaker has explicitly consented; do not use it to impersonate a third party.
SKILL.md
Loading
Related Openclaw Skills
NoMac: a Mac for your agent
NoMac gives your agent an administrator-enabled cloud Mac for macOS development, Xcode builds, iOS Simulator testing, and app delivery.
Website to Shopify Liquid Section
Convert a live website section into a complete, responsive Shopify theme section built with Liquid.
Website Section to Elementor Template
Convert a section of a live website into a styled, import-ready Elementor template for WordPress.
Clone Section
Clone Section extracts a live website section and turns its real styles and structure into code for your chosen framework or site builder.
QR Image Scanner
QR Image Scanner decodes QR codes and barcodes from local images while preserving original payload bytes, detection metadata, and scannability results.
LinkedIn Post
Create, preview, schedule, and publish first-person LinkedIn posts from an AI agent through PostWire’s approved integration—without browser automation or password sharing.
Featured*
