the_tuul
the_tuul copied to clipboard
Make a decent karaoke video from any song in about 10 minutes.
Currently, The Tuul uses a model that (tries to) leave in background vocals. Sometimes, it doesn't do a good job, and leaves in foreground vocals that it thinks are background...
Would it be possible to create a Docker version for easier deployment?
Example: ``` ע/רב_של_שו/ש/נים נ/צא_נא_אל_ה/בוס/תן מור_ב/ש/מים_ו/ל/בו/נה ל/ר/ג/ל/ך_מפ/תן ``` This shows up in the ass file with syllables going right to left, but the syllables are arranged LTR and the animation happens...
The code base is still missing a license, I recommend adding one
Video creation fails in Opera, on prod, with a `TypeError: failed to fetch` error printed to the console.
Add CDG export so videos can be played at actual karaoke joints.
When there's 3+ screens which follow up each other quickly, they could be splitted so bad that it would be impossible to tell where's part of which and what line...
Use [ffmpeg.wasm](https://github.com/ffmpegwasm/ffmpeg.wasm) to create the video in the frontend rather than on the server. The endgame is to have the whole tool run in the browser so that server costs...
[Jassub](https://github.com/ThaUnknown/jassub) seems better for subtitle previews. It does feel like the current version is a fraction of a second off from what the end result video ends up being.
Sometimes a screen will segue into the next screen very quickly and you don't have time to read what's on the next screen before it starts. Normal karaoke tracks get...