Hi Antigravity team,
I’ve been actively using the Antigravity IDE and wanted to share two pieces of feedback regarding the audio messaging feature:
1. Bug: Missing Chat Titles for Audio-Only Conversations Currently, if I start a new chat and only send audio messages, the IDE fails to automatically generate a chat title. The chat simply appears as an empty row (blank name) in the chat history. The title generation only triggers later if I manually type and send a text message. It seems the title generation module is only listening for keyboard text input events and ignores the transcribed text from audio messages.
2. Feature Request: Support for Custom Speech-to-Text Providers While the built-in Speech-to-Text engine is helpful, it occasionally struggles with non-English languages (like Bulgarian) and specific technical programming jargon. It would be incredibly powerful to have a setting or an extension mechanism that allows users to connect their own custom Speech-to-Text engines (for example, by plugging in an OpenAI Whisper or Deepgram API key). Having the ability to upgrade the transcription accuracy would make audio prompting much more reliable.
Thanks for building such a great tool!