Open 3ll3d00d opened 5 years ago
Is this a relatively trivial thing to do using ffmpeg, or are you planning to use another method? The main reason I ask is due to my observations with the full range signals and specifically in the 2-6kHz range, having an easy way to get some information about that specific range would probably be really helpful to me in addition to the R128 check I'm already doing.
This is just comparing filtered vs unfiltered FR in this range and changing the level of the original track by this much. Assumption is no filters are set that affect this range except for gain adjustments and the point of the exercise is to equalise the with/without overall volume level to make the comparison "fair" (or at least fairer)
ie this is not an overall normalisation exercise
Ahh, okay, I just saw that you were using 1-4kHz range and thought that since the 1000Hz sample rate files roll off well before that maybe you were doing something different.
I did realize this wasn't intended to be an overall normalization exercise, I just thought the same technique might be useful to give me an easy way to get data on a specific frequency range.
What sort of info are you after?
Peak data for a particular range, probably 2-6kHz.
It feels like this is a peak finding problem, i.e. show me the frequencies where a mix spikes away from the norm (where the norm is for that mix not some idealised norm like a loudness curve)
The way I am looking at that range of peaks is sort of as a ceiling on MV boost, which I can check in addition to R128.
compare FR in the 1-4kHz frequency range for filtered vs unfiltered, set offset accordingly