@Gabriela Macovei It sounds like the agent's voice recognition might need some fine-tuning to better understand Romanian, especially at normal speaking volumes and without the need for over-emphasizing words. Since the transcriber is currently set to English with the 'Nova 2' model, you'll need to switch to a model that supports Romanian and is possibly more sensitive to quieter speech. To address this issue, you should consider the following steps: 1. Ensure that the transcriber you're using supports the Romanian language. 2. If it does, switch the language setting from English to Romanian. 3. Test the transcriber with the new settings to see if it accurately captures "yes"/"no" and numbers in Romanian without the need for raised voices or over-emphasis. 4. If the issue persists, you may need to explore additional models or settings that are optimized for Romanian and can handle a variety of speech volumes and clarity. If you need assistance with changing the transcriber settings or finding a model that supports Romanian, please let me know, and I can guide you through the process. If the transcriber you're using doesn't support Romanian, you might need to look for an alternative solution that does.

I need assistance with changing the transcriber settings and finding a model that supports Romanian

Try another call by setting

smartEndpointingEnabled

to false.