I'm trying to handle interruptions and trasncribe the text that has been spoken before the interruption - how can I get the reverse mapping to text to be able to do that from the audio that was generated?
#Getting word timestamps using AsyncSSE TTS
vague moss
sleek notch
Hi @vague moss - sorry not too sure what you mean by reverse mapping to text - could you clarify? Maybe an example would help 😅
vague moss
Yes! For example, I am converting something like "Hello, world!" to audio, and, at some point when the audio is being played on the frontend, the user interrupts. I want to be able to transcribe the text afterwards. Did the user interrupt between "Hello" and "world"? Or after everything was said? Or before anything at all was said?
I can do this using the timestamps, but I then need to be able to map those timestamps back to the words (like word_timestamps in websocket), however I don't see this option in asyncsse