Audio and Video fingerprinting.

I went to VSF (Video Services Forum) conference in 2024 and talked to Paul Briscoe from TAG Video Systems. The topic was “Content Identification and Matching using SMPTE Standardized Fingerprints”. A few months later, VSF posted his presentation on YouTube, it’s worth watching.

SMPTE2064 introduces audio and video fingerprinting technologies, discusses how they’re done and how that information is carried in ISO13818-1 (MPEG-TS) transport containers, enabling downstream systems to assess lip sync and other interesting characteristics about the stream. The fingerprints help determine how LIP SYNC evolves throughout video workflows, helps designers and engineers understand if certain workflow components are creating unwanted drift. The fingerprints also let you quickly find content across mass live-stream video planes, super handy.

I talked with Paul, I basically said I was shocked that nobody had already implemented libraries.

Anyway, It took a while to find the time but I eventually changed that.

Here’s my Github Repo that implements the smpte2064 algorithms. The code is failthful to the spec, designed to be read along for easy debug/validation. It’s relatively easy to integrate and should limp up fairly quickly for any video s/w developer. It has obvious comments throughout where I think it can be optimized. Reach out to me if you have questions, otherwise - Enjoy!