Update on Diane - version 0.2

After running version 0.1 for a while now, it was time for a small update. It’s still a prototype, but since I’ve used it for real work a few things needed fixing. One of those was the length of recorded material, as about 10 minutes of recording wasn’t enough.

After doing a live demonstration of Diane where I talked for too long and she couldn’t handle the transcription, I tackled this.

Updated

Appearance / GUI I’ve switched from Tkinter to CustomTkinter for the GUI, for a slightly nicer look. :) Audio file conversion The recorded audio files are now converted to mp3, so you can record for 1.5 hours instead of about 10 minutes. No more template buttons To be able to add more templates, I’ve added a dropdown menu instead. GPT-3 vs GPT-4 Option to choose between GPT-3 and GPT-4. GPT-3 will be enough for most templates, but for more advanced ones you can choose GPT-4. Transcription only You can now also choose to just transcribe text without selecting a template. Some optimizations Some tweaks and optimizations of the features.

Code

You can find the updated code at github.com/mickekring/Diane

Questions?

If you have any questions, just ask :)