Can you get audio data from YouTube or Spotify?

Community Help
1

Is there any way to get the wave form data of a song playing without having to upload the mp3 file? In order of preference, it would be nice to:

1. Get the audio data playing on the computer speakers.
2. A screen share feature like in zoom, teams or discord, that gets the audio from another browser tab that is playing the music.
3. Copy the YouTube or Spotify link into observable and get the audio from the video playing in the notebook.
4. Run the YouTube or Spotify link through a third party site that extracts and send the audio file to the notebook

Are any of these possible?

The reason Iā€™m asking is Iā€™d like to make audio visualizer art, and having to upload an mp3 would make it too much of a hassle to use.

2

Have you seen this notebook?

1 Like
3

For 4 thereā€™s Yout API | Yout.com, although I havenā€™t tested it myself.

If you want to create your own solution that streams the audio on the fly Iā€™d maybe explore how youTube downloaders implement the download process, e.g. node-ytdl-core/lib/index.js at f5b57a2cdc87711454b4ff00bdad83363d42c668 Ā· fent/node-ytdl-core Ā· GitHub

1 Like
4

I havenā€™t seen that, but that is something Iā€™d definitely like to use to adjust my microphone before video calls ( :star:ed). Thatā€™s the kind of audio visual display I want to make, but it still doesnā€™t allow me to use music Iā€™m listening to as an input.

I played around with chat GPT and Claude for a few hours before posting in the forum, and the only solutions it gave me that worked was using the microphone as an input, or uploading an MP3 file to observables. The limitation is the security of web browsers in general, and additional security observable puts on the notebook. Its very understandable, because I donā€™t want hackers to listen to my computer audio if I have their site open in a tab. Additionally, YouTube uses iframes, so apparently it isnā€™t possible to extract the audio file just from the link.