openclaw - 💡(How to fix) Fix [Bug] Edge TTS reads SSML/XML markup as text instead of speaking content [1 comments, 2 participants]

TL;DR

The Edge TTS provider is not correctly parsing the SSML content-type, causing the XML tags to be read aloud instead of the intended text.

Guidance

• Verify that the Content-Type header is set to application/ssml+xml in the WebSocket request to ensure the Edge TTS server recognizes the SSML format.
• Check the Edge TTS server documentation to see if there are any specific requirements or limitations for handling SSML content.
• Consider modifying the ttsPromise method to validate the response from the Edge TTS server to ensure it is correctly parsing the SSML content.
• Test the Edge TTS provider with a simple SSML example to isolate the issue and determine if it is specific to the current implementation.

Example

// Example of a simple SSML request
const ssml = `
  <speak version="1.0">
    <voice name="zh-CN-XiaoxiaoNeural">
      <prosody rate="1" pitch="1" volume="1">
        Hello, world!
      </prosody>
    </voice>
  </speak>
`;
_wsConnect.send(`X-RequestId:...\r\nContent-Type:application/ssml+xml\r\nPath:ssml\r\n\r\n${ssml}`);

Notes

The issue may be specific to the Edge TTS provider or the current implementation, and further investigation is needed to determine the root cause.

Recommendation

Apply a workaround by modifying the ttsPromise method to remove or escape the XML tags before sending the request to the Edge TTS server, until the provider correctly supports SSML content-type.