I started using Gemini Live API for my Livekit cloud agent (model “gemini-2.5-flash-native-audio-preview-09-2025”). I appreciate the latency and low cost, but it seems to exhibit a “stutter” where it starts a word and is unable to finish it (see example). I have already confirmed that this is not an issue on livekit side. It seems that the audio being returned by google is garbled. Any suggestions or clarification would be appreciated.
GB_DEV
1
that’s so funny, thanks for the example!!
Have you met the issue with model 12-2025 ?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Gemini Live Model Regression | 2 | 182 | April 2, 2025 | |
| Facing some serious lag in responses with Gemini 2.0 in audio modality in multimodal live API | 2 | 302 | June 11, 2025 | |
| 2.5 flash audio native - output broken in DE | 8 | 627 | October 18, 2025 | |
| Gemini Live API: token generation suddenly stops | 13 | 1161 | October 8, 2025 | |
| Audio Chunks getting delayed (Stammering) from gemini flash 2.0 live 001 model in AI studio and in APIs as well | 3 | 173 | November 7, 2025 |