Inconsistent transcription language with gemini-2.5-flash-preview-native-audio-dialog

[bethuelpn]

Hello team,

We are experiencing an issue with the gemini-2.5-flash-preview-native-audio-dialog model regarding its transcription capabilities for the Portuguese language.

Issue or question in detail:
Our goal is to send audio in Brazilian Portuguese and receive a transcription in the same language. However, the model's behavior is inconsistent. Sometimes it correctly transcribes the Portuguese audio into Portuguese text, but at other times, it transcribes the same audio into a different language.

[marckevin]

Hi @bethuelpn,

Welcome to Google Cloud Community!

To ensure consistent audio transcription results, make sure the audio quality is high, with minimal background noise, and that the audio input remains consistent. Additionally, the gemini-live-2.5-flash-preview-native-audio model is currently in Preview, which means it is still under development, and may not yet offer the expected quality, and might have limited support. Preview features often come with certain limitations and might not provide the full range of quality or functionality available in the final product. It may have bugs or unexpected behaviors. However, you can expect the quality to improve as the feature matures.

Was this helpful? If so, please accept this answer as “Solution”. If you need additional assistance, reply here within 2 business days and I’ll be happy to help.