Studios

Speech recognition, synthesis, translation, speaker separation, turn detection and media transport are separate parts of a voice configuration. Latency is measured end to end for each deployment.

• Real-time speech transcription
• Asynchronous audio transcription
• Speaker diarisation
• Language detection
• Multilingual transcription
• Speech-to-text translation
• Speech-to-speech translation
• Text-to-speech output
• Pronunciation guidance
• Speech speed and tone settings
• Voice activity detection
• Turn detection and interruptions
• WebRTC transport
• WebSocket telephony transport
• Human call transfer
• Recording and transcript consent

The sample workspace does not open a microphone, place calls or generate audio. Recording and speech rights require consent.

Avatar presentation, bidirectional audio/video, camera interpretation and meeting participation each require their own session configuration.

• Interactive avatar sessions
• Bidirectional audio/video
• Avatar persona selection
• Avatar appearance studio
• Consented camera input
• Field-assistance visual review
• Document and object inspection
• Live video-call training
• Meeting transport evaluation
• Human video handoff
• Session consent and stop controls
• Avatar usage-rights review

Camera access must be explicit and revocable; the sample workspace does not start live avatar or camera sessions.

Creative requests define reference assets, output format, shot direction, brand constraints, rights and release approval.

• Image generation briefs
• Video generation briefs
• Reference-image direction
• Product campaign asset briefs
• Shot and camera direction
• Brand asset review
• Asset version and approval tracking
• Usage-rights review
• Campaign export planning
• Human publication approval

The sample workspace prepares briefs rather than generating or publishing media. Production media access is separate.