I want to make a compilation video of Tesla FSD Beta's disengagements. When FSD Beta disengages, it makes a particular sound. I would like a piece of software I can use without having to code that will show me the timestamps of each time the sound is heard. Ideally, it would be a video editor or something similar which could also allow to trim the video. I would prefer if it was free or at least free trial.


2 Answers 2


You'd probably want to download them, and process the audio, so that you could search faster than realtime. Doing that without a paid, premium subscription might violate YouTube's Terms of Service, and it might even violate the TOS WITH a premium subscription, so be sure you study those carefully, first.

If you can manage to download them, you might be able to use fast Fourier analysis to fingerprint the sound, and then search the files for it. You don't necessarily need to know all of the details about the signal processing, as long as you can find a tool that'll help you do it. I'd recommend having a look at this article as a starting point, and searching online code repositories like GitHub for something that suits your level of engagement and code familiarity. Even if you don't have any experience coding, you can sometimes find links to helpful google colabs or jupyter notebooks installations. Here's one I found one earlier today for music segmentation, that will analyze music from YouTube links and spit out Midi keyboard parts, just by clicking through the colab.

If you aren't able to download the files, you could still theoretically use software to analyze the stream of audio coming through your speakers, worst case, but you'll still need to find suitable tools to fingerprint it, and I'm less confident that you'd be able to find an off-the-shelf solution which works that way.

  • I was hoping for some software which just allow me to put in a sound file and find the timestamps after downloading but what you suggested sounds rather complicated. Anyway, thanks for putting the effort to help me. Feb 6, 2021 at 2:27
  • 1
    Honestly, I feel bad for posting such a crummy answer. The only reason I did is that I think it’s such an interesting question, and I hope somebody can help me help you find a better one. Anybody know if FFMPeg already has a built-in way to do this? @gyan? Feb 6, 2021 at 2:42

I'm not aware of any easy software to do that, but I'm posting anyway, since it could be useful to others as well.

There are open source projects that attempt to do what you require, basically by applying what @Jason Conrad said in his answer.

One of them is craigfrancis/audio-detect that you can find on GitHub. You first need to generate samples of the sound that you want to detect, and generally the more you have the better the performance of the algorithm will be. This can be done quite simply with the command that you find in the readme.

After that, the actual processing can be run, which takes the samples you extracted and tries to find instances of the same sound in an input file.

It's pretty easy to run both operations if you know a little about command line. Unfortunately, there's probably no magic solution for this task.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.