Hello, We would like to build a voice assistant and sell it in Quebec, Canada. For this province, it is a legal obligation that one provides services in French, if the client wishes. While we have seen that Vapi supports providers that offer transcribing and voice generation in French, we haven't been able to determine if we can support the following flow: A client calls the voice assistant's phone number, and starts speaking in English or French. Then, the voice assistant continues the conversation in the language that the client used. Could you please let us know if this flow is possible? If it isn't, do you plan to make it possible? Could you please offer us alternatives? Thank you, Radu

@radu Hello Radu, Yes, the flow you described is indeed possible with VAPI. Our multilingual support allows the voice assistant to detect the language being spoken by the client and respond in the same language. Here's how it can work: 1. The client calls the voice assistant's phone number. 2. The voice assistant uses the Transcribers (Speech-to-Text) feature to transcribe the spoken language. 3. The language detection can determine whether the client is speaking English or French. 4. The voice assistant then uses the Voice (Text-to-Speech) feature to continue the conversation in the detected language. This ensures a seamless experience for your clients in Quebec, complying with the legal requirements to provide services in French. If you need further assistance with setting up this flow or have any more questions, please let us know! Best regards, VAPI Support Team

set transcriber.languageDetectionEnabled to true use 11labs voice which supports both languages ask the model to generate response in language it recieves.

Not sure I understand, is this feature only available via the api or also via the dashboard? We tried to use deepgram multi but it did not work for italian

Vapi would benefit from creating Python / JS examples for calling APIs. Some key features are in specific API parameters and not in the UI - I'm also trying to wrap my head around it.

only the first has to be enabled using api. second is dependant on voice used. third has to be done using prompting.

Great, thanks for the info

Thank you for your response, do you have documentation for this?

for the first one you can update using vapi api docs. second you have to do your research and same goes for third according to your assistant.