Real time lyric detection

Setup

$ python3 -m pip install -r requirements.txt

Set the environment variables in .env or elsewhere:

• ACR_ACCESS_KEY
• ACR_ACCESS_SECRET
• SPOTIFY_ID
• SPOTIFY_SECRET

Then simply

$ python3 acrcloud_cmdline.py

The program will first record audio from the microphone, then identify it using acrcloud. Next the identified song is looked up on LRC sites

OR

$ python3 lyrics_spotify_camera.py

This program will find the spotify playing song, and create a camera using OBS-Camera with lyrics.

Current bugs with acrcloud_cmdline

• If you find a snippet in a chorus, the lyric timing will be off if it identifies the chorus in another part of the song instead.

• Sampling can lead to the program searching for a very old or relatively unknown song before the main song.

• The original swear words cannot be recovered if the word is completely starred out. E.g. **** does not work, but d**k -> dick.

Other quirks

The out.wav file and logs/res_dump.json are written to the current directory every record / identification cycle.