mdolr
commented
7 months ago Hey @katal1234 glad you enjoyed the app, to make sure you have a full context, this is more of a proof of concept I've built for myself over the week-end than a true competitor to rewind.ai
What database does this project use when recording continuous voice data?
Voice is not stored, the only thing stored is the transcript & screenshots metadata, the database used to store & query all of this is SQLite
How is the database query efficiency ensured as the data volume gradually increases?
it's not (I didn't test it for too long)
How is the scene restored during the search? (I believe this is one of the best features of software like Rewind.AI)
There is a slider along the timeline which you can move, once the slider stops the program will then display the closest screenshot in time as well as all the transcripts until the selected timestamp
Is there consideration for integrating with conversational large models to optimize the search experience?
I don't see how a conversational LLM would optimize search experience, perhaps you mean an embedding model, this could be something to do, I also wanted to add a conversational LLM to build an assistant on-top with context augmented search based on embeddings & queries asked but it's currently not the case
Please consider the project as archived, I'd suggest checking out https://github.com/jasonjmcghee/rem which is far more active!
Hi,Nice to meet you, and it's a pleasure to meet you and to see your open-source project, which has been very helpful to me. I have a few questions that I would like to inquire about.
Thank you very much! and Looking forward to your reply!