1. download tiktok video from url

uses yt_dlp library

2. extract tiktok audio as mp3

uses ffmpeg

3.  upload audio to suno
4.  cover audio

5.  get transcript with timestamps of tiktok audio/ get transcript with timestamps of suno audio
    uses replicate
    villesau/whisper-timestamped:c5b122b7e513b1b5a6ef849891c538869b77cc932cbd0f8203e11d3b357553b8 https://replicate.com/villesau/whisper-timestamped/versions/c5b122b7e513b1b5a6ef849891c538869b77cc932cbd0f8203e11d3b357553b8?prediction=wkn0tthczxrm80crc3p9zx0agw
6.  compare
7.  combine song with suno audio and use speed ramp tecnique to sync them

            make video with moviepy

    todo

8.  get transcripts for tiktok audio in background while uploading to suno
9.  if no anchor points found when comparing transcription use openai api to find commonalities in transcripts
10. speed up the whole process -can we use the streaming url? does compressing our tiktok audio help?
11. ux improvements: start downloading tiktok vid straight away and then show prompt to make the user feel the generation is faster/ Show hooks while the user waits.
12. find a better way to do comparison:
    Victor linked https://chatgpt.com/share/688ae631-3070-8000-9a36-a31dcdb373c6

    a) look at aligned_lyrics endpoint?
    b) Does midi or other data available to us help?

13. can we generate lots of songs at once in parallel? we could then show them in a tiktok style ui
14. skip long intros