TT-DR dynamic range measures: cannot replicate; ideas on calculation? 2011-09-01 16:00:23 Hello DearsI am new in this forum. I am not an audio eng. But I have quite a good math/computer background and an insane passion for music. Last week me and other 5 mate started to compare several mastering (same record mastered by different audio eng.) in blind testing and then we wanted to see whether the TT-DR measures (http://www.dynamicrange.de/) correlated with our rakings. For instance we compared Desperado/Eagles (I have four different masterings), Abraxas/Santana (3 different masterings), Waltz for Debby/Bill Evans Trio (4 different masterings) and many others. In most cases DR failed to replicate our rankings (which by the way was almost always the same for all of us). Only in few instances some of my buddies agreed with DR meter. So I started to play around with it. I read the manual and I wrote a small MATLAB script to compute the TT-DR following what they say here:www.pleasurizemusic.com/sites/default/files/DR-Manual-V1_1-English.pdfTo be honest I got crazy but I cannot replicate their numbers... eiher I didn't understood what they do or they did not write the documentation accurate enough. Does somebody know how precisely they compute those numbers?Best RegardsPierre

TT-DR dynamic range measures: cannot replicate; ideas on calculation? Reply #1 – 2011-09-01 19:38:12 My read is that you break the file into 300 microseconds long pieces and do an RMS measurement on each, create a histogram from these measurements, discard the lower 80% of the measurements and average the remaining 20%. Compute the peak level for the entire file and subtract this from the average you computed to get the DR value.I'm curious, what criteria were you using to compare the masters?If you did not normalize loudness before auditioning, you can expect the louder master (lowest DR) will almost always win.

TT-DR dynamic range measures: cannot replicate; ideas on calculation? Reply #2 – 2011-09-01 20:10:36 I'm not really qualified to comment, but I'll throw-out a couple of thoughts....For comparing the dynamics of different masters of the same recording, I'd just compare the average (or RMS) to the peak. This approach won't work when comparing different-unrelated recordings, but it should tell you which copy of the same recording is more compressed.Quotethis result, onlythe loudest 20% is used for determining the averageloudness of the loud passages.At the same time, the loudest peak is determined.The DR Value is the difference between the peak and the top20 average RMS measurements (top 20 RMS minus Peak =DR). Do I understand correctly that they are ignoring the quietest 80% of the recording, and comparing the loud passages to the loudest peak? That doesn't make sense to me... But like I said, I'm not qualified... I haven't done any experiments or studied the algorithms. And, they are using the loudest peak as a reference? One short-loud "transient" peak might throw-off the calculations without making the song sound more dynamic overall. Quote... eiher I didn't understood what they do or they did not write the documentation accurate enough.It's probably those RMS calculations... They talk about measuring the RMS value over the whole song, and they talk about making 10000 measurements??? And, the 10000 measurements thing is strange... You'd normally make RMS calculations over a fixed period of time, (i.e. every 10ms, etc.) rather than splitting the song into 10000 equal parts.... if that's what they are doing??? It might not be possible to directly calculate the RMS value over the entire song... I think the sum-of-squares number may become unmanagable. It think it's normally done by calculating the RMS values of short time-segments, and then averaging those RMS values.

TT-DR dynamic range measures: cannot replicate; ideas on calculation? Reply #3 – 2011-09-01 21:21:03 Quote from: Notat on 2011-09-01 19:38:12My read is that you break the file into 300 microseconds long pieces and do an RMS measurement on each, create a histogram from these measurements, discard the lower 80% of the measurements and average the remaining 20%. Compute the peak level for the entire file and subtract this from the average you computed to get the DR value.I did this even though the *histogram* part is a bit puzzling. Why they need to construct an histogram why not taking a sample quantile at 80%? And there is another variable which is important here which overlapping between windows.Quote from: Notat on 2011-09-01 19:38:12I'm curious, what criteria were you using to compare the masters?If you did not normalize loudness before auditioning, you can expect the louder master (lowest DR) will almost always win.No we used a calibrated SPL meter to adjust the volume pot in order to have the same average level at listening position.Best Pierre

TT-DR dynamic range measures: cannot replicate; ideas on calculation? Reply #4 – 2011-09-01 21:26:24 Quote from: DVDdoug on 2011-09-01 20:10:36...For comparing the dynamics of different masters of the same recording, I'd just compare the average (or RMS) to the peak. This approach won't work when comparing different-unrelated recordings, but it should tell you which copy of the same recording is more compressed.NO, average cannot work. The mean is biased by few large squared samples... it's not robust at all. Quote from: DVDdoug on 2011-09-01 20:10:36It might not be possible to directly calculate the RMS value over the entire song... I think the sum-of-squares number may become unmanagable. It think it's normally done by calculating the RMS values of short time-segments, and then averaging those RMS values.This is not an issue, you can easily compute such RMS. The point for windowing is that dynamic can only be captured as a local measure rather than global.BestPierre