espnet VS yt-dlp

You might check out this list from espnet. They list the different corpuses they use to train their models sorted by language and task (ASR, TTS etc):

https://github.com/espnet/espnet/blob/master/egs2/README.md

- https://github.com/espnet/espnet

This link has instructions for doing this for a Japanese model. It would have to be altered to work with ljspeech and the fine tune dataset.

This work is made possible by the excellent advancements in text to speech modeling. ESPnet is a great project and should be checked out for more advanced and a wider range of use cases. This pipeline was also made possible by the great work from espnet_onnx in building a framework to export models to ONNX.

CMU WavLab has ESPNet https://espnet.github.io/espnet/ which includes a number of high quality TTS models including VITS (which in my subjective experience is just as good as what is demonstrated here). Also the inference on various ESPNet pretrained TTS models is reasonable and sentences take on average 5 seconds per word to generate the waveform on my totally mid PC setup.

The reason I'm saying this is to point out that having and in-depth knowledge on speech processing/generation requires a lot of information about signal processing and human speech in general (eg. acoustics and phonetics). However, if you're not into learning everything there is to know about a subject, just take one state-of-the-art example and study that as best as you can. Pick one environment/toolkit, for example espnet and simply go with that.

https://github.com/espnet/espnet (kind of like a newer Kaldi, but also not beginner friendly)

both provide e2e ASR support but espnet does have more utilities where as speechbarain is clean

This is relatively small amount of speech to train the model from scratch, but you can train using another pre-trained model for initialization. There are numbers of end-to-end ASR toolkits which can be used for this: https://github.com/NVIDIA/NeMo and https://github.com/espnet/espnet

You can put these options in a config file and they will become the default: https://github.com/yt-dlp/yt-dlp?tab=readme-ov-file#configur...

Yep. yt-dlp and youtube-dl

https://github.com/yt-dlp/yt-dlp

https://github.com/ytdl-org/youtube-dl

Will also start to feel the impact. My theory is that we will see a bunch of new video hosting sites as youtube itself attempts to lock down its ecosystem. They haven't paid attention in any adversarial way as far as I can tell.

When they do, it wont be great.

The points you make aren't unreasonable.

It is necessary to establish clear boundaries of what can and can be provided by the maintainers. If not done at an earlier stage of the project, the support burden becomes too much to bear at which point the maintainer transfers ownership, and the project suffers from catastrophic consequences such as the xz backdoor we're talking about here, or other cases where the project mostly stalls and serves as an ego-boosting platform for the new maintainer, as was the case with PhantomJS[6].

This can also happen in your life, where a "friend" sees that you possess a certain skill, and then gradually tries to push an inordinate amount of their personal work related to this field onto you.

Personally, I think it's best to use an approach with extremely clear communication as to what the maintainer can and cannot provide. This can be seen, for example, in yt-dlp[1], where the consumer is clearly informed upfront that not providing detailed information as requested will lead them to block said consumer; or sqlite where their position regarding contributed patches[2] and support[3] is similarly made clear.

Having a shouty BDFL like Torvalds can also help improve code quality[4] and questionable contributions[5], though it is better that the shouty BDFL makes statements that are professional and do not show as much aggression; so for example, "Mauro, shut the fuck up"[7] would become "Mauro, your response is completely unbecoming for a Linux kernel maintainer, and is not in line with the promise of not breaking userspace."

[1] https://github.com/yt-dlp/yt-dlp/issues/new?assignees=&label...

[2] https://www.sqlite.org/copyright.html

[3] https://www.sqlite.org/support.html

[4] https://www.theregister.com/2024/01/29/linux_6_8_rc2/

[5] https://cse.umn.edu/cs/linux-incident

[6] https://github.com/ariya/phantomjs/issues/14541

[7] https://lkml.org/lkml/2012/12/23/75

Or just "yt-dlp "

yt-dlp ( https://github.com/yt-dlp/yt-dlp ) still works pretty well at the current state of Twitter.

To download audio from YouTube videos, you'll utilize the widely used yt-dlp library, which can be installed using the pip command as follows:

Use: `yt-dlp with --write-comments --no-download --batch-file FILE`

- FILE is a text file with a list of YouTube id's/URL's

- https://superuser.com/a/1732443/4390

- https://github.com/yt-dlp/yt-dlp

> Then youtube-dl wasn't a thing anymore (maybe it is again?)...

yt-dlp is definitely a thing: <https://github.com/yt-dlp/yt-dlp>