Voice Mode

Voice mode lets you speak naturally to the AI, making sessions feel more like conversations than typing. Your speech is transcribed in real time with fast, accurate speech-to-text that handles technical terminology and code references well.

Using Voice Mode

In any session, activate voice mode from the input area. Hold the space bar (or tap the microphone button) to record your message. Release to send. The AI will respond with text and optionally with audio.

How to use

• In any session, activate voice mode from the input area.
• Hold the space bar (or tap the microphone button) to record your message.
• Release to send. Your speech is transcribed and sent as a message.
• The AI responds with text and optionally with audio.

Transcription

Speech-to-text is fast and accurate, even with technical terminology and code references. Voice input is transcribed using Deepgram's speech-to-text API, providing reliable transcription of technical discussions.

Audio Responses

Audio responses use high-quality text-to-speech that can be toggled on or off per session. Toggle this in session settings based on your preference.

Multi-User Sessions

Voice mode is particularly natural in multi-user sessions where the team is essentially having a meeting with an AI participant. Everyone can speak and the AI responds as part of the conversation.

Cost Tracking

Speech-to-text and text-to-speech usage is tracked against your credit balance. You can monitor voice-related costs alongside your regular session usage.

Speech-to-text is inexpensive. Text-to-speech is more costly due to high-quality voice synthesis. Toggle TTS off in session settings if you prefer text-only responses.