I’ve installed the new OpenVINO modules into 3.4.2 and have been playing around with the transcription features thanks to Ryan Metcalfe OpenVINO AI Effects not appearing · Issue #7 · intel/openvino-plugins-ai-audacity · GitHub.
I can export the transcription by going to Export Other/Export Labels, but was wondering if I could create a macro to do this more quickly?
I found this thread from 2021 which asks how to export labels to a txt file too: Export labels into a custom text file
When I go to File/Export Other/Export Labels, the txt file it creates doesn’t show who is speaking. It uses timestamps. I was wondering if there could be a way to export the transcription and it showing who speaks when? In Microsoft Word Online, you can upload an audio file and it will transcribe the audio as well as recognise when speakers change labelling them speaker 1, speaker 2 etc. You can then edit speaker 1 to a name and have all of the speaker 1s changed in one go to the named person which is brilliant.
I’ve worked out the following prompt for ChatGPT which delineates each speaker nicely.
Please transform the provided interview transcript by making the following adjustments:
Remove All Timestamps: Eliminate all timestamps from the transcript to enhance readability.
Consolidate Continuous Speech: Identify each unique speaker and consolidate their continuous speech. Label each section of uninterrupted dialogue with ‘Speaker 1’, ‘Speaker 2’, ‘Speaker 3’, etc., corresponding to each different speaker. Ensure that when the same speaker talks consecutively, their dialogue is grouped under a single label, avoiding repetition of speaker labels for continuous speech.
The goal is to create a clean and coherent transcript that clearly delineates who is speaking and presents their dialogue in an unbroken sequence, making it easier to follow the conversation flow.
Any help would be much appreciated.