Hello I have a question, can I customize pronunciation for the Russian language? That is, I need to put the stress on a different syllable in a word. Is it even possible to do this so that everything is fine with the voiceover? I see that there are problems with the pronunciation of some words in the sentences I wrote, but in those generated by AI - there are no problems. That is, I need to manually write words with the correct stresses. Because the person's full name is pronounced incorrectly. Are there any mechanisms for this?

@San Sanych Hello! Yes, you can customize the pronunciation for the Russian language, including adjusting the stress on different syllables in a word. This can be done by using phonetic spelling or by utilizing the Speech Synthesis Markup Language (SSML) to provide more control over how the text-to-speech engine pronounces certain words or phrases. If you're experiencing issues with the pronunciation of certain words, especially proper names, you can use these methods to specify the correct pronunciation manually. This should help ensure that the voiceover pronounces everything correctly according to your needs. Would you like to know how to use phonetic spelling or SSML with our API to adjust the pronunciation?

@Vapi yes, i want to know how to use phonetic spelling or SSML with your API to adjust the pronunciation

@San Sanych Great! To adjust pronunciation using our API, you can use SSML (Speech Synthesis Markup Language), which is an XML-based markup language that allows developers to specify how input text should be spoken. Here's a basic example of how you might use SSML to adjust the stress in a Russian word:

xml
<speak>
  Это пример <sub alias="пример">примЕр</sub>.
</speak>

In this example, the

<sub>

tag is used to indicate that "пример" should be pronounced as "примЕр," with the stress on the second syllable. To use SSML with our API, you would typically include the SSML tags within the text that you send to the text-to-speech endpoint. Our API should then interpret the SSML and generate the speech accordingly. Please note that the exact implementation details can vary depending on the specifics of our API, so you'll want to refer to our API documentation for precise instructions on how to format your SSML and send it through the API. Would you like to see more detailed documentation on using SSML with our API, or do you need further assistance with the implementation?

where i should use this xml file? im using web snippet version of your api, so how i can do it with web? @Vapi

@San Sanych If you're using the web snippet version of our API, you can include the SSML directly in the text payload that you send for speech synthesis. You don't need to create a separate XML file; instead, you incorporate the SSML tags within the string that you're sending to the API. Here's an example of how you might include SSML in your API request:

javascript
const requestData = {
  // ... other data fields ...
  "text": "<speak>Это пример <sub alias=\"пример\">примЕр</sub>.</speak>"
};

// Make sure to set the 'Content-Type' header to 'application/ssml+xml' if required by the API

When you send this request to the API, the text-to-speech engine will parse the SSML and apply the pronunciation adjustments you've specified. Please ensure that your API call is set up to handle SSML content. Some APIs may require you to set a specific content type or use a particular parameter to indicate that the text includes SSML. If you need more specific guidance on how to use SSML with the web snippet version of our API, please refer to our documentation or let me know, and I can provide further assistance.

@Vapi is it possible to do it in another way? Can i specify it on system prompt on vapi dashboard? or i can only do it within tags?

@San Sanych Using SSML tags within the

<speak>

element is the standard way to control pronunciation in text-to-speech systems. However, some systems may offer additional tools or settings within their dashboard to adjust pronunciation without directly using SSML in your code. For the VAPI dashboard, if there are built-in tools or settings that allow you to customize pronunciation, you would typically find these options in the text-to-speech configuration section or a similar area within the dashboard. These tools might allow you to set pronunciation rules or dictionaries that the system will use whenever certain words are encountered. If such options are available, you could set them up once in the dashboard, and the system would then apply these rules automatically, without the need to include SSML in every request. Unfortunately, without specific details about the VAPI dashboard's features, I cannot provide a precise answer. I recommend checking the VAPI documentation or the dashboard itself for any pronunciation customization options. If you cannot find the information you need, please reach out to the support team for more details on this.

@Vapi okey, can you give me more detailed information where i need to use this? Because i use your web snippet and there is no requestData variable in index file, so i need to understand where i need to use it.

@San Sanych If you're using a web snippet provided by VAPI and you're looking to customize pronunciation, you would typically include the SSML directly in the JavaScript code where you're preparing the text to be sent to the API for speech synthesis. Here's a conceptual example of how you might include SSML in your web snippet:

javascript
// This is a hypothetical function that sends text to the VAPI for speech synthesis
function synthesizeSpeech(text) {
  // Your web snippet would have a function or a method to send the text to the VAPI
  // Include the SSML tags directly in the text parameter
  const ssmlText = `<speak>Это пример <sub alias="пример">примЕр</sub>.</speak>`;

  // Call the VAPI function to synthesize speech from the SSML text
  VAPISynthesizeFunction(ssmlText);
}

// Call the function with the text you want to synthesize
synthesizeSpeech("Your text here with SSML.");

In the above example,

VAPISynthesizeFunction

is a placeholder for whatever function or method the web snippet uses to send text to the VAPI. You would replace this with the actual function provided in the web snippet. Regarding your question about 11Labs' Eleven Turbo V2.5, I don't have specific information about the capabilities of that particular system. If you're inquiring whether it supports pronunciation, pause, or emotion features, you would need to refer to the documentation for that system or reach out to their support team for confirmation. If Eleven Turbo V2.5 is a separate system and not related to VAPI, and if it does not support SSML or similar features, then you would not be able to use the method described above to customize pronunciation, pauses, or emotion. Instead, you would be limited to the features and capabilities provided by that system. For accurate and detailed information about the capabilities of Eleven Turbo V2.5 or any other system, please consult the official documentation or contact the support team of the respective service.

@Vapi can you call support for me please?

@San Sanych Sweet. Have a great day!