Voice input
Speak your messages instead of typing them, and choose whether the transcript lands in the composer or sends straight away.
Speaking to Panda
Voice input lets you dictate messages instead of typing them. Click the mic button in the composer to start listening, and PandaOS transcribes your speech. The mic stays open until you stop it, so you are not holding a button down while you think.
There are two ways to finish. Click the mic again and the transcript lands in the message field, where you can read it, edit it and add to it before sending. Press Send while it is still listening and PandaOS transcribes and sends in one go, without the detour through the composer.
Voice works well for longer, free-form requests where typing would slow you down: explaining a bug you are looking at, describing what you want built, or thinking through an approach out loud. Everything else - the model, interaction mode, attachments - works exactly the same whether you type or speak.