Speak replies out loud
Tell an OpenClaw agent to use ClawVoice from now on. It calls the CLI for user-facing replies while still writing the full answer in chat.
OpenClaw voice skillClawVoice gives your OpenClaw Agent a voice.
Spoken replies, terminal push-to-talk, hosted x402 fallback, local voice setup, and a small-balance Base USDC wallet built for OpenClaw agents.
ClawVoice CLI
$clawvoice init --mode hybrid --mic
$clawvoice install-mic
$clawvoice talk
$clawvoice stop
Agent instruction
Use ClawVoice to speak your responses out loud from now on. Stop talking when I say stop.
62
first-day installs observed
31
language codes in the global guide
x402
pay-per-call voice pricing
Base
USDC wallet network
What it does
A voice layer for agent workflows.
Talk back with push-to-talk
Run install-mic once, then clawvoice talk. Speak normally, press Enter when done, and ClawVoice transcribes locally before the agent answers.
Use local or hosted voice
Use a free local voice engine after setup, hosted x402 voice when local install is not available, or hybrid mode for local-first fallback.
Keep spend controlled
A local Base USDC hot wallet, approval prompts, daily caps, session caps, balance checks, and withdrawals keep hosted calls bounded.
Cost model comparison
Local voice, hosted x402 voice, and cloud TTS are different buying models.
ClawVoice is optimized for OpenClaw agents: local-first when possible, hosted x402 fallback when needed, and wallet-level spend controls. The package signs the endpoint's x402 challenge directly; it does not add a separate hidden ForgeMesh fee. The comparison below uses public pricing pages checked on July 3, 2026; provider prices can change.
ProviderBuying modelPublished cost signalBest fit
ClawVoice local
Free after local setup
No per-call hosted voice charge
OpenClaw users who can install the local runtime and want private, repeat use.
ClawVoice hosted x402
pricing sourcePay per approved call from a local Base USDC wallet
Endpoint-declared x402 price; base voice is currently $0.001 per successful hosted call
Agents that need hosted fallback without a monthly voice subscription or hidden add-on fee.
ElevenLabs
pricing sourceSubscription credit pool
Free 10k credits; Starter $6/mo with 30k credits; Creator $22/mo with 121k credits
Creator workflows, voice cloning, studio tools, and broad audio production.
Google Chirp 3 HD
pricing sourcePer-character Google Cloud billing
$30 per 1M characters; other Google TTS tiers range from $4 to $160 per 1M characters
Google Cloud teams that want managed TTS with cloud billing and quota controls.
Amazon Polly
pricing sourcePer-character AWS billing
Standard $4, Neural $16, Generative $30, Long-Form $100 per 1M characters
AWS-native applications that already use IAM, CloudWatch, and AWS billing.
Azure Speech
pricing sourcePer-character Azure billing, region and tier dependent
Free tier includes 0.5M neural characters per month; paid pricing varies by region and tier
Microsoft/Azure environments that need enterprise speech services and procurement controls.
Customize your agent's voice
Use the OpenClaw page for agent voice. Use voice.forgemesh.io for the hosted voice surface.
This page is OpenClaw-specific: install, setup, ClawHub, wallet behavior, push-to-talk, and the CLI options an agent needs. The hosted voice endpoint can keep its own API-focused surface at voice.forgemesh.io.
Voice ID
M1, F1, or hosted voice IDs
Language
31-language guide, default en
Endpoint tier
base, pro, custom, long variants
Preset
hosted preset pass-through
Mix
hosted mix pass-through
Expression
expression plus level
Expression controls
name=value controls
Global-friendly
Built-in guide for 31 language codes.
Language support is exposed through clawvoice voice --lang. The hosted service can support additional aliases, while the page and CLI give users a clear global starting point.
Language codes
enesfrdeitptnlplcshuroelsvdafinoruuktrarhehibnidmsfilvithjakozh
FAQ
Answers agents and users can quote.
What is ClawVoice?
ClawVoice is an OpenClaw and ClawHub skill from ForgeMesh Labs that gives agents spoken replies, terminal push-to-talk input, hosted or local voice output, and a local Base USDC x402 wallet with spend controls.
How do I make my OpenClaw agent speak out loud?
Install and initialize ClawVoice, then tell the agent: Use ClawVoice to speak your responses out loud from now on. The agent should call clawvoice speak for normal user-facing replies.
How do I talk to my agent with a microphone?
Run clawvoice install-mic once, then run clawvoice talk. In the current terminal push-to-talk flow, speak normally and press Enter when you are done talking.
Can ClawVoice use hosted voice if local voice is not installed?
Yes. In hosted or hybrid mode, ClawVoice can use hosted ForgeMesh Voice through x402 when the local wallet has a small USDC balance on Base and the user approves paid calls.
Can I customize my agent voice?
Yes. ClawVoice supports voice ID, language, endpoint tier, preset, mix, expression, expression level, and expression controls through setup or the clawvoice voice command.
How does ClawVoice pricing compare with hosted TTS providers?
ClawVoice local mode has no per-call hosted voice charge after setup. Hosted ClawVoice uses approved x402 calls from the local wallet, while providers such as ElevenLabs, Google Cloud, Amazon Polly, and Azure Speech usually bill through subscription credits or cloud per-character pricing.
Install
Install from the release bundle or ClawHub, then run clawvoice init.
Fund hosted fallback
Send a small USDC balance on Base when using paid hosted voice.
Test the loop
Say hello, stop talking, then restart voice mode to verify the full agent behavior.