Does anyone know a private way to transcribe video/audio? It doesn’t need to be perfect, I just don’t want that data saved with some tech company.
#asknostr
Discussion
Use whisper locally
What is that?
And open source speech to text model from meta. You can run it locally on videos, audio etc. Check for example
https://github.com/lablab-ai/How-to-use-Whisper-to-transcribe-a-YouTube-video-Tutorial
I once ran a DVM doing it for people but there was no interest in it.
I don’t want to upload to YouTube though 🧐
You don't need to. The first part is to download from YouTube and then transcribe. If you already have the video, skip the first part. Just feed in the path of the local file
So this all done locally? Open AI doesn’t get shit from me?
On android or PC? Download whisper from openai github for PC. FUTO voice input for android
PC or Ubuntu. Can you trust an open AI product?
If you don't trust it, unplug your PC when using it. Yes, I mean Linux. Why would you use anything else?
It'll work on either and it's open source https://github.com/openai/whisper.
There's also a cpp whisper implementation (not openai) which is prob more optimised for running on a CPU. Bit more finicky to set up though, you need to download the models you want to use then build it, etc. Instructions are straight forward though.
https://github.com/ggml-org/whisper.cpp
nostr:nprofile1qqs0w2xeumnsfq6cuuynpaw2vjcfwacdnzwvmp59flnp3mdfez3czpspremhxue69uhkummnw3ez6ur4vgh8wetvd3hhyer9wghxuet59uq3samnwvaz7tmwdaehgu3wd3hhqupwwdhkx6tpdshslw6x5e had a good wrte up on Whisper https://blog.lopp.net/openai-whisper-transcription-testing/
No idea