Is there any way I could implement Google STT in my agent. Since I am using a foreign language that urgently needs Google STT to be ready for production

@Esrever Absolutely, integrating Google Speech-to-Text (STT) into your agent can be a great way to handle foreign language processing. To implement Google STT, you'll need to use the Google Cloud Speech-to-Text API. Here's a high-level overview of the steps you would follow: 1. **Set up a Google Cloud account**: If you haven't already, you'll need to create a Google Cloud account and set up a project. 2. **Enable the Speech-to-Text API**: In your Google Cloud project, you'll need to enable the Speech-to-Text API. 3. **Authentication**: You'll need to authenticate your application with Google Cloud, typically by setting up a service account and obtaining a JSON key file. 4. **Install the Google Cloud client library**: For the programming language you're using, you'll need to install the appropriate Google Cloud client library. 5. **Implement the API**: Using the client library, you'll write code to send audio data to the Speech-to-Text API and receive the transcribed text. 6. **Handle the response**: Once you receive the transcribed text, you'll integrate it into your agent's workflow. 7. **Testing and optimization**: Test the integration thoroughly and optimize for the specific characteristics of the foreign language you're working with. If you need more detailed instructions or assistance with the code, please let me know which programming language you're using and I can provide more specific guidance. If you encounter any issues or have further questions, don't hesitate to ask for support.

Yes, please give me more details. Can I do this inside of Make.com?

@Esrever Certainly! Make.com (formerly Integromat) is a versatile automation platform that allows you to connect various services without writing code. To integrate Google Speech-to-Text (STT) with your agent through Make.com, you can follow these steps: 1. **Google Cloud Setup**: Before you start on Make.com, ensure you've set up a Google Cloud account, enabled the Speech-to-Text API, and obtained the necessary credentials as mentioned in the previous message. 2. **Make.com Scenario**: Log in to your Make.com account and create a new scenario. 3. **Google Services Module**: Look for the Google Cloud Speech module in Make.com's list of services. If it's not available, you can use the HTTP module to make a custom API request to Google STT. 4. **Configure the Module**: If using the Google Cloud Speech module, follow the prompts to connect your Google account and configure the module with the correct language and audio settings. If using the HTTP module, set up the API request according to Google's API documentation, including the endpoint, headers, and body with your audio file and API key. 5. **Process the Response**: After setting up the module to send audio to Google STT, you'll receive a response with the transcribed text. You can then add additional modules to process this text within your agent's workflow. 6. **Test Your Scenario**: Run the scenario to ensure that the audio file is being sent correctly and that you're receiving the expected transcribed text. 7. **Error Handling**: Implement error handling to manage any potential issues with the audio transcription process. If you need more specific instructions or run into any issues, please let me know, and I'll be happy to assist further. Make.com has a user-friendly interface, but sometimes setting up API requests can require a bit of trial and error to get everything working smoothly.

@Vapi so how do I send the transcribed text into Vapi with a json when I have a transient assistant?

@Esrever To send the transcribed text into VAPI with JSON when you have a transient assistant, you would typically follow these steps: 1. **Obtain Transcribed Text**: First, ensure that you have the transcribed text from Google STT or any other transcription service ready in a variable or accessible format. 2. **Format the JSON Payload**: Create a JSON payload that includes the transcribed text in the format expected by VAPI. The exact structure of this JSON will depend on the VAPI's expected input parameters. 3. **Send the JSON to VAPI**: Use an HTTP request to send the JSON payload to VAPI's endpoint. If you're using Make.com, you can use the HTTP module to configure the request method (POST, GET, etc.), add headers, and include the JSON payload in the body. 4. **Handle VAPI's Response**: After sending the request, handle the response from VAPI. This could include checking for success, parsing any returned data, and taking subsequent actions based on the response. 5. **Integrate with Transient Assistant**: If the transient assistant is part of your VAPI setup, ensure that the assistant's state is managed correctly, and the transcribed text is passed to it in a way that allows the assistant to process it as intended. If you need to know the specific API endpoint or JSON structure for VAPI or have any other questions about integrating with a transient assistant, please provide more details or ask for support, and I'll be glad to assist you further.

@Vapi how would you add the transcribed text to this transient assistant instead of using deepgram as the provider? { "assistant": { "name": "{{10.

Name - Just used for squads

}}", "voice": { "voiceId": "{{10.voiceID}}", "provider": "{{10.

Voice Provider

}}", "speed": {{10.speed}} }, "model": { "model": "{{10.

GPT Version (model)

}}", "messages": [ { "role": "system", "content": {{9.json}} } ], "provider": "openai", "maxTokens": 250, "temperature": {{10.Temperature}}, "tools": [] }, "firstMessage": "{{10.

First assistant message

}}", "voicemailMessage": "You've reached our voicemail. Please leave a message after the beep, and we'll get back to you as soon as possible.", "endCallFunctionEnabled": true, "endCallMessage": "{{10.endCallMessage}}", "transcriber": { "model": "nova-2-phonecall", "language": "en", "provider": "deepgram" }, "serverMessages": [ "end-of-call-report", "function-call" ], "dialKeypadFunctionEnabled": false, "endCallPhrases": [ "Goodbye" ], "hipaaEnabled": {{10.hipaaEnabled}}, "recordingEnabled": {{10.recordingEnabled}}, "silenceTimeoutSeconds": 30, "responseDelaySeconds": {{10.responseDelaySeconds}}, "llmRequestDelaySeconds": {{10.llmRequestDelaySeconds}}, "llmRequestNonPunctuatedDelaySeconds": {{10.llmRequestNonPunctuatedDelaySeconds}}, "numWordsToInterruptAssistant": {{10.numWordsToInterruptAssistant}}, "maxDurationSeconds": {{10.maxDurationSeconds}} } }

You cannot use of google STT as of now.

Is it on the roadmap?

Yes it will be integrated soon.

Perfect! Cant wait!

You have to wait for sometime 🙂

Couple of months?

will let you know.

Any update on this?

no updates on this, and no ETA.

@Shubham Bajaj any update now?

same will update you till then you can raise request over https://vapi.canny.io