OPEN SOURCE - We read something that made us angry. Then we coded a solution.
Obviously, the officers’ goal was to hinder activists from uploading video footage to social media platforms. Every time Instagram, Facebook, or Youtube detect a piece of copyrighted audio, it blocks the video from being uploaded or shared. Within days, we researched and coded a tool to help activist against this new trend: unfilteredtruth.cc
Obviously, the officers’ goal was to hinder activists from uploading video footage to social media platforms. Every time Instagram, Facebook, or Youtube detect a piece of copyrighted audio, it blocks the video from being uploaded or shared. Within days, we researched and coded a tool to help activist against this new trend: unfilteredtruth.cc
The web app
unfilteredtruth is a fast and easy way to keep this kind of footage shareable, even if copyrighted music is being played in the background. It modifies the audio portion of a video to make it invisible to the social media copyright filters. To human ears, in contrast, the change is almost unnoticeable.If you want to post video footage of police misconduct on any of the platforms, it is fast and easy to handle. It’s accessible from desktop and mobile devices with a lot of browsers. After conversion, it lets you check the result.
The web app
unfilteredtruth is a fast and easy way to keep this kind of footage shareable, even if copyrighted music is being played in the background. It modifies the audio portion of a video to make it invisible to the social media copyright filters. To human ears, in contrast, the change is almost unnoticeable.If you want to post video footage of police misconduct on any of the platforms, it is fast and easy to handle. It’s accessible from desktop and mobile devices with a lot of browsers. After conversion, it lets you check the result.
Filters against filters.
Dedicated filters isolate the audio signal from the video. For easier comparison, audio waveforms are condensed into smaller chunks of data highlighting characteristic features. Algorithms like Fast Fourier Transformation, Mel Spectrogram, Tempogram, or Chromagrams calculate different aspects of the same waveform. Other algorithms look for unique characteristics – this process is called audio fingerprinting. Just as in real life, these fingerprints are compared with a database with fingerprints of original, copyrighted material.
Bypassing the filters
The original audio is rendered upon a very low carrier frequency. The effect is efficient enough to fool content filters. To make things even harder for the filters, we change the audio speed a little, almost unnoticeable to the human ear.
Safety for important data
Handling video proof can be a delicate topic. Thus, we refrained from sending video and audio to external servers. All data is kept safely on your computer and processed in your browser. To be totally transparent and open for improvements, we published the source code as open source under a very permissible license. Everybody is invited to double-check the repository on github and improve the code.
Bypassing the filters
The original audio is rendered upon a very low carrier frequency. The effect is efficient enough to fool content filters. To make things even harder for the filters, we change the audio speed a little, almost unnoticeable to the human ear.
Safety for important data
Handling video proof can be a delicate topic. Thus, we refrained from sending video and audio to external servers. All data is kept safely on your computer and processed in your browser. To be totally transparent and open for improvements, we published the source code as open source under a very permissible license. Everybody is invited to double-check the repository on github and improve the code.