I'm using Azure TTS Viseme events.
When using Azure tts viseme SDK, the audio offset of viseme seems to be wrong.
Some viseme comes before the real audio and some viseme comes after the real audio.
In start part of every sentence is more critical to use.
I attach a image file that summarizes the viseme problems that have been identified so far.
Is there any way to align these viseme audio offset to real audio?