Question about fps & recognition

[worldveil]

Question about fps & recognition

[worldveil]

1. Do you mean frames per second? Visual frames per second have no bearing on how many different samples the audio track emits per second.
2. No, it's the same algorithm.

[skrew]

Yes it's frames per second. I have tested and i confirm the recognition don't works when the FPS are not the same.

[worldveil]

You need to clarify, which is it:

1. http://en.wikipedia.org/wiki/Frame_rate (video)
2. http://en.wikipedia.org/wiki/Sampling_(signal_processing)#Sampling_rate (audio)

Audio and video sequences are separate signals, and the speed at which pixels flash across the screen has no bearing on the sampling rate of the audio, which is what dejavu concerns itself with.

Furthermore, you should be able to downsample (or upsample) audio before fingerprinting to a uniform sample rate like 44100Hz and have only moderate performance hits.

[skrew]

Yes, Frames per second (video). I confirm DVD / TV channels don't use the same FPS. As workaround, i test with differents ratios on audio, so i can even identify if source come from DVD or TV Channels. A bad news can become more functionalities :)