Oasis Browser: Composer Microphone and Speech-to-Text

Voice in an AI browser only helps if it fits how you already work: quick dictation into the composer, not always a full hands-free session.

Insight for other users: Oasis composer microphone is push-to-talk in the text box. Hold or click the mic, speak, then release. Your audio is sent for transcription, and the text lands in the composer field so you can edit before you send. This is a separate path from the full voice orb hands-free loop (listen, think, respond).

How to use it:

• Grant microphone permission when the browser prompts.

• Fix any wrong words in the transcript, then send with Enter like a normal message.

• Need a full voice session instead? Use the voice orb flow (different UX, same sign-in expectations).

Sign-in: Voice transcription uses Kahana cloud endpoints and expects you to be signed in. Guests see a sign-in prompt instead of results.

Limits: Poor network can drop uploads. Watch for inline errors and retry.

Doc: https://kahana.co/docs/composer-microphone
Related: https://kahana.co/docs/voice-orb-hands-free | https://kahana.co/docs/composer-and-follow-ups | https://kahana.co/docs/voice-and-sign-in

Questions we would love your take on:

1. Push-to-talk vs orb
   When do you reach for composer mic vs opening the voice orb? Should Oasis make that choice clearer?

2. Edit before send
   Is landing text in the composer (instead of auto-sending) the right default for privacy and accuracy?

3. Permissions and trust
   What would make you comfortable granting mic access in a privacy-first browser?

4. Errors and retries
   How should Oasis surface failed uploads on bad Wi-Fi?

5. For the Oasis team
   Did you discover the mic in the composer on your own, or did you assume voice was orb-only?

Share your voice workflow. We are improving both paths.