Currently, our TTS pipeline does not support timestamping. This is useful for various use cases like book readers, etc. Also, this has been asked by a community member on Discord. This task aims to check out this model and see if we can integrate a similar approach.