Waveform analysis in Kdenlive 2020

 Check this out!

Ok, so this is working pretty well. Remember, synchronization of video or audio clips based on audio waveform analysis  only works if you have good sound tracks on every recorders... unlikely the case if you're filming multicam from a flying GoPro... Maybe I'll drop smpte ltc generation from the workflow and synthesize Kdenlive optimized sound tracks? It's back to square one... almost!  8-)

I'm wondering if I could generate the sound track on the SAMD21 itself. Then no needs for batch processing after that... zoom, directly into Kdenlive! I can explore this possibility... I'll study Kdenlive algorithm, essentially an FFT convolution. Yes, "Read the Source, Luke" 8-).

If you want an explainer on correlation and FFT convolution, here's a good one from the California Institute of Technology.