Open hanimayahi opened 7 years ago
Sweet!
This makes me think, there must be something like OpenCV but for audio? Researching a little I find these tools: STK https://ccrma.stanford.edu/software/stk/ or AudioKit http://audiokit.io/
I believe it should be possible to test small sound segments of each transcodes compared to the source file. Because if the audio is offsync, the sound won't be similar at the exact same times between the two files.
If this really works we can completely automate the QC process.
oh snap.. if that works out... game changer..
After #34 is done, opencv (https://github.com/peterbraden/node-opencv) can be used to automate the similarity between the transcode and source files. Then we won't need a human to check that the episodes are correct and FPS synced, only to check audio sync.
If the opencv check returns a high probability rate, then it's with most certainty the correct source.