Voice and transcription

Say what is on your mind instead of typing. Voice input turns speech into text you can review before sending. Read aloud lets you listen to an answer. Open your profile menu in the sidebar, then choose Settings → Voice.

​

Voice mode

Turn this on to show recording and transcription controls. Turn it off to hide microphone controls and stop an active recording. This is separate from listening to answers: turning voice input off does not turn read-aloud off. In Chat, choose Record voice beside the message box. Allow microphone access if your browser asks. Review the transcribed text before sending, especially names, dates, measurements, and medication doses.

​

Microphone

Choose System default to use your device’s default microphone, or select a specific microphone. This helps when you have both a headset and a built-in microphone. Under Microphone access, choose Test and speak. Moving bars show that the app is receiving sound. The test runs in the browser and does not send a chat message. Choose Stop when finished. Leaving Voice settings also stops the test. Not checked means you have not tested it. Connected means the test is active. For Permission blocked, check browser and device microphone permissions. For No microphone found, reconnect the microphone or select another. Browser or device limitations may also prevent a test.

​

Clean up transcripts

Request punctuation and grammar cleanup after transcription. Cleanup is intended to preserve meaning and medical details, not add an answer or new information. Check the result: speech recognition and cleanup can make mistakes.

​

Transcription dictionary

Add words you often say, such as an unfamiliar name or specialist term. These give transcription spelling hints; they do not guarantee recognition or automatically replace words. Enter a term and choose the add button. Duplicate terms are ignored. Limits are 100 terms, 50 characters per term, and 2,048 characters in total. Remove terms you no longer need. Your dictionary, cleanup preference, and recent transcripts are stored encrypted for your account in this browser. Do not assume they will appear on another device.

​

Recent transcripts

Find your latest completed transcriptions here. The list keeps up to 20 entries; it is not a permanent archive. Save important information in the appropriate conversation or your own records. Clearing browser data can affect locally stored information.

​

Read responses aloud

Turn this on to hear new assistant answers when they finish. To listen to one answer, choose Read aloud on that message. Your browser may require an interaction before audio can play. Enabling this setting reveals the following controls.

​

Playback speed

Choose 0.75×, 1×, 1.25×, 1.5×, or 2×. Slower playback gives you more time to follow the answer. It does not change its words.

​

Speech engine

Browser TTS uses voices available through your browser and device. Speak Aloud API uses the app’s cloud speech service and needs a connection. Browser voices vary by device; not every voice is guaranteed to work offline.

​

Language and Voice

These appear with Browser TTS. Language filters available voices; it does not translate the answer. Voice selects who reads it. Changing Language clears the previous voice selection so you can choose a matching voice.

​

Enable sound effects

Play a short cue when recording starts and stops. This is separate from reading answers aloud.

​

Use enhanced audio engine

Ask the browser to apply echo cancellation, noise suppression, and automatic microphone gain. These can help with background noise and speaker echo. Results depend on your browser and microphone.

​

Dynamic normalization

Smooth sudden microphone-volume changes before audio reaches transcription. This can help when you move closer to or farther from the microphone. It changes audio levels, not transcript wording.

​

Get help

Check your selected microphone and permissions for recording problems. For playback, check device volume and output. For a repeatable problem, contact us and choose Report a bug. Include your browser, the setting involved, and any error reference. Leave private recordings and health details out unless you intend to share them with support.