| from services.gemini_client import get_gemini_client | |
| from google.genai import types | |
| async def speech_to_text(audio_bytes: bytes) -> str: | |
| client = get_gemini_client() | |
| # Wrap audio bytes correctly | |
| audio_file = types.File(data=audio_bytes, mime_type="audio/wav") | |
| response = client.models.generate_content( | |
| model="gemini-2.5-flash", | |
| contents=[audio_file] | |
| ) | |
| return response.text | |