@0xbudhdhiman It's great to hear that you're utilizing VAPI with such a comprehensive configuration! You're correct that the voice is generated by Eleven Labs and the transcription is likely provided by Deepgram. The reason for the discrepancy between the voice and the transcript could be due to the transcription service interpreting the audio, which can introduce errors. To achieve 100% accuracy between the voice and the transcript, you would indeed need to use the text generated by OpenAI as the source for both the voice synthesis and the subtitles. This would bypass the need for a separate transcription step. However, this functionality would depend on whether VAPI and the integrated services support such a feature. If VAPI has a feature to directly use the text from the model for subtitles, you should enable that in your configuration. If this feature is not available, you might need to request it from the VAPI development team or check the documentation for any existing options that you might not be aware of. For now, if you want to explore this further or if this feature is not currently supported, you could reach out to VAPI support for more information on how to achieve this. They might provide you with a workaround or inform you about upcoming features that could solve this issue.

resolved