search

met4citizen / TalkingHead

Talking Head (3D): A JavaScript class for real-time lip-sync using Ready Player Me full-body 3D avatars.
MIT License
349 stars 107 forks source link

Issue with some voices with google text to speech #73

Closed audreytimbani closed 5 days ago

audreytimbani commented 5 days ago

Voice : en-GB-Studio-B

Error :


{
  "error": {
    "code": 400,
    "message": "`\u003cmark\u003e` tags are not currently supported by Studio voices.",
    "status": "INVALID_ARGUMENT"
  }
}
audreytimbani commented 5 days ago

Got it

Error Explanation: The error message states that " tags are not currently supported by Studio voices," leading to an INVALID_ARGUMENT status. This suggests that while you are using valid SSML syntax, the specific voice you selected does not support this feature .

Feature Limitation: As of now, Studio voices, despite being generally available since February 26, 2024, do not implement support for tags. Users have reported similar issues and have suggested filing a feature request for this capability

met4citizen commented 5 days ago

Yes, this is a known issue. From the TalkingHead README: "The used [Google] voice must support SSML and \ tags that are needed to get word-level timestamps. Currently, Google supports SSML and \ tags when using Standard, Wavenet, Neural2, News, or Casual voice types."