Speak, review, send.
Choose a chief, press Dictate and allow microphone access. Press Finish recording to transcribe. The words appear in your editable message box; sending is always your choice. Escape discards a recording.
TALK IT THROUGH
Your words.
A clearer first draft.
Use Dictate in your chief chat to record up to 30 seconds. Review the transcript before Send.
Hear the reply.
Use Read reply after a response completes for Orpheus English playback. Spoken output is brief; the full reply stays in chat. Stop reading at any time. Choose Talk live · English for a voice conversation: speech is sent after a pause, your chief replies using Orpheus English, and listening resumes after playback. End voice or Escape stops the loop. This uses consecutive spoken turns, rather than simultaneous speech.
Whisper and Orpheus, within your allowance.
Transcription uses production Whisper models available to our Groq account. Spoken replies use the Orpheus English preview model when active. Preview availability can change; if speech fails, text stays available. An active trial or subscription is required. Recording length is checked on the server, and speech inference uses your compute budget. Raw microphone audio is not saved to chat history; a transcript enters history only when you send it.
If your microphone is unavailable.
Allow microphone access in browser settings and use a browser with AudioWorklet support over HTTPS. If recording or transcription fails, your existing draft stays available and you can type instead.
Our revenue is ₹5 lakh per month, expenses are ₹7 lakh and cash is ₹24 lakh. Calculate runway, explain assumptions and suggest three practical options.