#1245660177953194066 What should I do to have the assistant read the tool messages in the same language as the one being used in the conversation? So, if Spanish is being spoken, the content is read in Spanish (not English). Exemple: "messages": [ { "type": "request-start", "content": "Give me a second while I check that..." }, { "type": "request-failed", "content": "Sorry. I was not able to retrieve that information. Would you want the listen the regular hours instead?" }, { "type": "request-response-delayed", "content": "We're experiencing a brief delay communicating with our server.", "timingMilliseconds": 4000 } ],

@Jonas To have the assistant read the tool messages in the same language as the one being used in the conversation, you would need to implement a system that detects the language of the conversation and selects the appropriate voice model for text-to-speech (TTS) output in that language. Here's a step-by-step approach: 1. **Language Detection**: Use a language detection feature to determine the language of the user's input. 2. **Language-Specific Voice Model**: Choose a TTS provider that supports multiple languages, such as Eleven Labs or Azure. Configure your system to select the voice model that matches the detected language. 3. **Translation (if necessary)**: If the tool messages are not already in the target language, you may need to translate them before using TTS. 4. **Text-to-Speech Output**: Use the TTS service with the selected voice model to read the messages aloud in the detected language. For example, if the conversation is in Spanish, the system would: - Detect that the language is Spanish. - Select a Spanish voice model from the TTS provider. - Read the messages using the Spanish voice model. If you need to implement this feature or have any questions about how to integrate multilingual support into your voice assistant, please let me know, and I'll be happy to assist you further!

change the content to spanish langauge.

but this is a multilingual prompt

sorry for the delayed response somehow this got missed. the messages array content is of forced type so they are said in language they are written.