The idea was to test Gladia and its multilingual ASR model "accurate". It works but it is very slow and results in high latencies. I created a squad and used it for the main agent as a language detection step. Then each agent of the squad is in a specific language, with a Deepgram ASR model. The main issue is latency. Even with language agents. Scenarios are short, no knowledge base. Do you have any idea recommendation? Thanks

Hey @zamzam To help track down this issue, could you share: - The call ID - When exactly this happened (the timestamp) - What response you expected to get - What response you actually got instead This would really help me figure out what went wrong!

@Shubham Bajaj thanks for your reply. call id: 3c77c0b8-0f0c-4d43-b91d-17c14a80335a timestamp: from the very beginning .response expected: transcription in French (french is being spoken) , response i get: text translated and transcribed in english instead of French, from the very begining-> as a consequence, language is not being recognized and conversation not transmitted to the right agent. it used to work, although it was slow, I am using this code for the main member of the squad: "transcriber": { "provider": "gladia", "languageBehaviour": "automatic multiple languages", "model": "accurate" } thanks!

@zamzam did you update the first assistant of your squad?

@zamzam I'm trying to do exactly what you tried here - with openai realtime as first agent, to then dispatch in the different languages. Have you been able to do so ?

@Jeebs i think that s a good use case. Although i did not tested it. i am not sure realtime api can work well with squads, we will have to test.

Thanks for your answer ! I'm honestly struggling with both squads and blocks 😢

@Jeebs from eva.be Sorry for the late response. Are you still experiencing this problem? If so, could you share the following in a new #1211483291191083018 ticket: - The recent call ID - When exactly this happened (the timestamp) - What response you expected to get - What response you actually got instead This would really help me figure out what went wrong!