tl;dr: This [website](https://freesubtitles.ai/) offers a free interface to openai’s whisper, which generates automatic transcriptions at very high quality. To use it, use “Japanese” in the language field and “small”, “medium” or “large” in the model field.
I’ve been playing around with whisper myself for a while now and the quality of the transcriptions is unmatched (but not perfect). I think this is a large help for practicing listening comprehension, you can e.g. apply this to podcasts (probably also an interesting option for the creators). For those interested, I applied it to [this video](https://www.youtube.com/watch?v=4DryV3xN-M0), here’s the [transcription](https://drive.google.com/file/d/1X2gkXU5BXmFpYtdLbgROxXCvfjTFqsjO) (I added [spaced kanji](https://drive.google.com/file/d/12cxNxo3I1a0PIU4L-X43dVnZmEvcCvux), [spaced hiragana](https://drive.google.com/file/d/1U5mzMe8aYgyh_mQuP2uvAieN5SYHJR58) and [roumaji](https://drive.google.com/file/d/1kKO7Rihq-7ESRPLf7SY2L0aZy_c7LYRe) captions for those who can’t fluently read “real” Japanese yet. This adds additional errors, please only use this for practicing listening comprehension together with the audio, NOT for isolated reading practice. You can get the roumaji captions also from google translate).
Whisper has been [posted here already](https://www.reddit.com/r/LearnJapanese/comments/xljc3e/whisper_a_new_free_ai_model_from_openai_that_can/), but only with the option of using it via google colab. The interface of the site above is better for casual use. I got this website from [HN](https://news.ycombinator.com/item?id=33663486), in case you are interested in some technical details. I think this is a generous offer from the creator, this probably won’t be around forever as it costs money…
Last but not least: What is said in the video also applies to learning Japanese. Don’t overdo it and have fun :).