6 ms·
Next: 1. Prompt it to extract the audio track, then give it to a speech-to-text API, translate it to another language, then make it add it back to the video fi
by sho_hn 4y ago
Next:
1. Prompt it to extract the audio track, then give it to a speech-to-text API, translate it to another language, then make it add it back to the video file as a subtitle track.
2. Retrain the model to where it does this implcitly when you say "hey can you add Portuguese subtitles to this for me"?
- stnmtn 4y agoI don't have words for how much this seems like a relatively trivial thing to do now, and 1 year ago I would have laughed at someone if they suggested this was a possibility in 5 years I'm feeling a mixture of feelings that I can't begin to describe
- iamwil 4y ago"Falling forward into a future unknown"
- sho_hn 4y agoThe Star Trek computer is here! :-)
- JaDogg 4y agoBut not the Star Trek economy / elimination of money. So some of us will starve while we get there. :(
- TechnicolorByte 4y agoWell said. It so easy to take for granted all these tech milestones with generative AI in particular the last year.
- stevenhuang 4y agono retraining may be necessary, this is a common enough ffmpeg task I wouldn't be surprised it can do it right now as a one-shot prompt. what a time to be alive!
- swyx 4y agoholding on to my papers fr fr
- ManuelKiessling 4y agoYou mean like this? https://app.fyyn.io/en/features#element-10 https://app.fyyn.io/en/features#element-10