Skip to content

matiaszanolli/devinyl

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

41 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

DEVINYL - Recover vinyls beyond recovering

DEVINYL is a tool with a very simple purpose in mind: Restore vinyls. The older and more damaged they are, the better the work it does.

How does it work?

Contrary to v1, DeVinyl v2 is as simple as it can get. Most vinyl records have at least a ~2 second audio gap before actually starting each song, specially old 78rpm records. So taking a sample from seconds 1 to 2 you can be pretty sure you're getting a pure noise sample (I'll make this adjustable in a future release for your custom tracks). From there, by using the awesome open source (SoX)[https://sourceforge.net/projects/sox/] library, we can create a noise profile from the source track and with said profile, just remove the majority of the noise of the song, including heavy hissing and clicking and without creating noticeable artifacts.

Requirements

  • Any modern OS
  • SoX binaries (available as packages on Linux via Apt / dnf / pacman, must download from website in Windows / macOS: (https://sourceforge.net/projects/sox/)[https://sourceforge.net/projects/sox/])
  • ffmpeg binaries (same as above, though available in windows through chocolatey / Winget and macOS through brew)
  • (Windows only) check that SoX and ffmpeg are accessible through terminal, otherwise add their respective location to your user's PATH environment variable.

Usage

Linux/macOS

./devinyl.sh source_file

Windows (Powershell)

./devinyl.ps1 source_file

The clean track will be generated as <source_track>_clean.flac

Extras

Additional reference sites

These links helped me a lot better understanding the topics of audio processing and noise reduction in general:

Noise reduction gist

Noise reduction topic

noisereduce library

matchering

matchering-cli

Anaconda

FFmpeg

If this helped you out, please buy me a coffee!

https://www.buymeacoffee.com/matiaszanolli

Or follow my channel in YouTube:

https://www.youtube.com/@TechforMusicAI