Skip to main content
Voice chat is an optional capability that your administrator must enable after configuring its voice service. In a new or existing main chat, select the blue waveform button. Allow microphone access when your browser asks. The composer moves to the bottom immediately. A short chime signals that voice is ready; then speak naturally. Hubbla replies aloud and adds captions to the same conversation. You can interrupt while it speaks. The spinning end button indicates that voice is still connecting; select it to cancel. Use Mute microphone to pause your input and End voice conversation to return to typing. Ending voice immediately stops local microphone capture and playback. If playback is blocked, select Play assistant audio. Voice uses OpenAI to process microphone audio and generate speech. Hubbla saves text captions and task status in your chat; this feature does not save audio recordings. Captions may differ from what you said or heard. Check names, dates and important results. Requests that need workspace information or actions go to Hubbla’s existing task backend. A task card shows its status and result separately from the spoken reply. Ending voice does not cancel accepted tasks; use Stop task to request cancellation. Actions already completed cannot be undone by cancellation. When you correct a running task, review its status before resubmitting the revised request. Existing approval steps still apply. Voice is available in ordinary main chats; page Preview and page voice notes keep their existing behavior. Sessions end after two minutes of inactivity or a maximum of fifteen minutes. If the connection to Hubbla stops responding, voice stops and displays recovery guidance. You can continue by typing or start a new voice session when ready. Saved captions and accepted task state remain available. If microphone permission was denied, update your browser permission and retry. Cancelling before your first spoken message does not create an empty chat.