Dictations
Dictations enable voice input directly in the chat. Via the microphone icon in the chat input field, you start voice input; your spoken words are converted to text using Speech-to-Text and passed to the AI agent. This is especially useful when you want to formulate quickly and freely without typing.
Difference from other audio features
Dictations are not recordings that are saved and transcribed later. Instead, the voice input is immediately converted to text and used directly in the chat – similar to typing a message, only spoken. In contrast, in-person meetings and online meetings capture entire conversations that are saved as audio files, transcribed, and further processed by AI agents.
How to dictate
Click the microphone icon in the chat input field and speak your text. The AI converts your words into text in real time. Once you are done, you can review the text before sending and adjust it if necessary.
Advanced settings
Under User Preferences you can customise speech recognition:
- Transcription language – set the language for Speech-to-Text or use automatic detection
- Review before sending – review the transcribed text before sending
- AI cleanup – remove filler words and correct grammar
- Swiss spelling – replace ß with ss
- Transcription prompt – optional prompt to control style and vocabulary (max. 750 characters), e.g. Formal meeting notes with technical terms
- Word replacements – automatically replace words the model misrecognises
For more information on using the chat, see Core Features.