I am using the whisper API to generate transcripts - but because the audio length is so long I have to chunk it. The problem is, when the transcript is reassembled, the timestamps are off. How do I fix this? Is there an option in the API to specify a duration offset at which the timestamps can begin?
Hi and welcome to the developer forum!
At the moment you need to open up the transcript file and modify the timestamp’s yourself with an offset that you keep track of, your code knows how long each segment is and can therefor add an offset to each of the transcriptions to account for that.