Voice Input

Deutsch: Spracheingabe

Instead of typing, you can dictate: AI Studio records your voice, has it transcribed, and puts the text on your clipboard. You then paste it wherever you need it, in the chat, in an assistant, or in any other program. AI Studio can also transcribe audio and video files that you attach.

What you need

A transcription provider, which turns speech into text:

1. Add one in "Settings" → "Configure Transcription Providers" with "Add transcription provider". It can run in the cloud or on your own computer; see Providers and Models.
2. Choose it in "Settings" → "App Options" → "Select a transcription provider".

Once a transcription provider is chosen, a microphone appears at the bottom of the navigation bar.

Dictating

1. Click the microphone, "Start recording your voice for a transcription". A short sound tells you that the recording runs, and the microphone turns into a red stop button.
2. Speak.
3. Click the stop button, "Stop recording and start transcription". A second sound follows.
4. AI Studio sends the recording to your transcription provider. When the text is ready, a third sound plays, and AI Studio copies the text to your clipboard.
5. Paste the text, for example, with Ctrl+V or Cmd+V.

AI Studio deletes the recording once the transcription is done.

On macOS, the system asks once for access to your microphone. Allow it, otherwise AI Studio can't record.

A keyboard shortcut

With a shortcut, you start and stop the recording from anywhere, even when AI Studio isn't in front:

1. Open "Settings" → "App Options" → "Voice recording shortcut", and click "Change shortcut".
2. Press the key combination you want. It must include Ctrl or Cmd, Shift, or Alt.
3. Click "Save".

To remove the shortcut, click "Clear Shortcut", then "Save". On Linux, the desktop may ask you to confirm the shortcut. If it can't register it, the shortcut only works while AI Studio is in front.

Audio quality

"Transcription audio quality" decides how well AI Studio compresses the recording before uploading it. Higher quality can improve the transcription of quiet or noisy recordings, but the upload gets larger. "128 kbps (recommended)" is the default.

Transcribing files

You can also attach recordings, audio files, and video files to a chat. AI Studio transcribes them and attaches the text; see Attachments.

When something goes wrong

• "The transcription result is empty.": the recording held no speech, or the microphone recorded nothing. Check your microphone in the settings of your operating system.
• "Failed to start audio recording.": AI Studio couldn't use the microphone, for example, because access was denied.
• The provider couldn't read the audio: AI Studio sends audio in the WebM/Opus format, which some providers may not accept. Choose another transcription provider.

Where to go next

• Attachments: transcribe recordings and videos in the chat.