DeepSpace voice
Voice is an optional DeepSpace input and output path. Its controls work only when the deployment has the realtime service and voice agent configured and the browser grants microphone access.
Speech input
Use the microphone control to dictate into the DeepSpace composer during a connected voice session.
Spoken response
Text-to-speech can speak a completed assistant response when the voice session and agent support it; it is not a promise of continuous spoken progress updates.
Text remains available
DeepSpace text chat does not require voice. If the realtime service is unavailable, voice may be disabled while text workflows remain separate.
Browser permissions
Microphone access requires a secure browser context and explicit permission. You can revoke that permission in browser settings.
Deployment requirements
Operators need to deploy and configure the realtime transport (such as LiveKit) and voice agent, provide secure HTTPS/WSS connectivity, and validate network routing including TURN where needed. The production Compose stack does not itself prove those external services are available.
Privacy and availability
Voice sends audio through the configured realtime service and agent. Review the privacy information for the deployment and any external voice provider before use. Local UI code or a simulated test does not establish physical-device, network, or hosted-production readiness.