Mulno never builds a face template, never builds a voice template, and never tries to detect a tester's emotions. A playtest's webcam and microphone streams, where a tester turns them on, are recorded as ordinary video and audio and watched or listened to by people — your reviewers, or (only under the conditions in the AI statement) a text transcript a speech-to- text provider produces.
Specifically:
- The webcam preview shown to a tester before they answer the consent screen exists only to let them see themselves; it is never recorded, and nothing analyses it.
- Our speech-to-text integrations (OpenAI, ElevenLabs) are sent only the parameters needed for a plain transcript with word timestamps — never a request for speaker diarization or a voiceprint.
- No automated system scores a tester's face or voice for identity, age, mood or attention.
This is a statement about what Mulno's own software and integrations do. A studio that adds its own analysis of the recordings it receives is outside Mulno's control and this statement.