Sign in with Google before you let the wolves loose.
🐺 OpenClaw Voice
Pick an agent, start talking
Checking login...
🎤
You
Waiting for voice...
00:00
Swipe down handle or tap ✕ to close
Mute
End call
Vision
Attach
New session
Tap if your phone blocks speaker playback
Voice Lock
Speaker verification
Keep calls tied to your voice.
Voice Lock compares incoming microphone audio with your trained voice track before the agent transcribes it. Once a profile exists for your account, it is used automatically for your calls. If no profile exists, the portal uses the normal voice flow.
Profile
Checking...
Mode
Checking...
Automatic use
Checking...
Failure policy
Fail-open when unavailable
How training works
1Use a quiet room and keep this device close, the same way you would during a call.
2The portal records seven simple samples and sends them to the existing Voice Lock enrollment endpoint.
3The server stores a voice profile for your signed-in account. You can retrain later to replace the current track.
Shadow and fail-open behavior
In shadow mode, Voice Lock scores audio but does not block transcription. When the verifier is disabled, missing, or unavailable, calls continue without blocking so you are not locked out by a service issue.
How calls behave
There is no per-call checkbox. If this account has a trained voice track, the server adds Voice Lock to new calls automatically. If there is no trained profile, calls continue through the normal microphone and transcription flow.
Train voice track
Training records seven simple samples. Read each prompt out loud; browser speech recognition advances after it hears the phrase, with Finished reading as a fallback.
Practice text
Hello, this is Mehdi. I am speaking in my normal voice for the voice track.
The text stays here while you read. It moves on when speech recognition hears enough of the phrase, or when you tap Finished reading.