Hello,
Like many, I discovered AI interaction via ChatGPT, then I also used Whisper from Open.ai.
I have some knowledge of using the linux shell, and some other basic linux knowledge. So my level is utltra beginner. Far from many of you.
I’m in the process of taking an EN-US language Youtube course of hundreds/thousands of hours of videos. I’ve been using yt-dlp+Whisper+deepl a lot, to try and make it easier.
I’m French and my listening/oral English isn’t very good. So I tried several approaches.
I wanted to do STT ----> Translation ---->TTS but I didn’t succeed. It’s not easy to keep a correct timestamp after translation. Or, since the Whisper transcription is of better quality than the Youtube one, I’d like to be able to at least subtitle them properly in French.
If you have any ideas or advice, I’d love to hear from you.
Could you give me some advice?
I saw that Meta with Seamless +M4T was going to offer a powerful tool, but we’re currently limited to a few seconds/minutes of STS + translation. Am I wrong?
I also have another question, but that may be another subject :
The project may be impossible for my level and the amount of computing power required, but…
Let’s imagine thousands of hours of restranscribed training videos, if I inject them into an AI, That I train this AI, then I can ask it any question on the various subjects, is that feasible?
Also, how do you start learning about AI? The subject is so vast !
Thank you in advance for your answers.