Hey Blue Cat users
I'm getting really inconsistent numbers with crest factor measurements. Is there anything in settings I can adjust to improve accuracy/repeatability, or is it just the nature of the measurement itself that is causing this?
Thank you
Meter Pro crest factor wildly inconsistent
-
- KVRer
- Topic Starter
- 9 posts since 7 Jun, 2013
If anyone is interested, if I let a clip loop through 3 times it greatly improves the consistency of the numbers.
With RMS average set @ 600 ms and no looping of the clip, crest factor ranged anywhere for 6.3 to 8.1. Pretty wide swing there @ 600 ms.
With looping of the clip 3 times, crest factor ranged from 7.0 to 7.2. Much improved and not surprising that these numbers are about the same as averaging the numbers from the non-looped test.
With RMS average set @ 600 ms and no looping of the clip, crest factor ranged anywhere for 6.3 to 8.1. Pretty wide swing there @ 600 ms.
With looping of the clip 3 times, crest factor ranged from 7.0 to 7.2. Much improved and not surprising that these numbers are about the same as averaging the numbers from the non-looped test.
-
- KVRist
- 355 posts since 30 Jun, 2009 from moon
deleted
Last edited by replicant X on Tue Mar 26, 2024 12:18 am, edited 1 time in total.
Speedrum 2 is a hidden gem. 
-
Blue Cat Audio Blue Cat Audio https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=39981 - KVRAF
- 6350 posts since 8 Sep, 2004 from Paris (France)
Hi,
Not sure about what you mean by "inconsistent". The crest factor is a dynamic value, so it may indeed change over time based on what has been played in the past. You may indeed want to adjust the RMS average time to change the way the crest factor reacts to changes.
Also, usually the histogram is more interesting than the value itself, as it gives you an overview of the distribution of values in the song.
Hope this clarifies!
Not sure about what you mean by "inconsistent". The crest factor is a dynamic value, so it may indeed change over time based on what has been played in the past. You may indeed want to adjust the RMS average time to change the way the crest factor reacts to changes.
Also, usually the histogram is more interesting than the value itself, as it gives you an overview of the distribution of values in the song.
Hope this clarifies!
-
- KVRer
- Topic Starter
- 9 posts since 7 Jun, 2013
Thanks for the responses.
I've run more successful null tests than I can count, so I'm not sure what is meant by DAW randomness.
By inconsistent, I mean that I run a short 4-bar piano clip through DP Meter Pro and record it's measurements. I then reset the values and run the same clip through again. The max peak/RMS measurements are always exactly the same, the average peak/RMS are around the same, but the average crest factor is never the same and many times not even close. I even tried comparing the crest factor of a clean run vs. sending it through an L2 limiter with the threshold pulled down as low as it can go, and the squashed version sometimes reported HIGHER crest factor than the non-limited one. I don't see how DAW randomness could be the reason for this considering the max measurements always end up the same and the clips always pass the null test. Voxengo SPAN also measures crest factor with about the same level of inconsistency. That's why I was wondering if it was just the nature of the measurement itself that causes this.
I've run more successful null tests than I can count, so I'm not sure what is meant by DAW randomness.
By inconsistent, I mean that I run a short 4-bar piano clip through DP Meter Pro and record it's measurements. I then reset the values and run the same clip through again. The max peak/RMS measurements are always exactly the same, the average peak/RMS are around the same, but the average crest factor is never the same and many times not even close. I even tried comparing the crest factor of a clean run vs. sending it through an L2 limiter with the threshold pulled down as low as it can go, and the squashed version sometimes reported HIGHER crest factor than the non-limited one. I don't see how DAW randomness could be the reason for this considering the max measurements always end up the same and the clips always pass the null test. Voxengo SPAN also measures crest factor with about the same level of inconsistency. That's why I was wondering if it was just the nature of the measurement itself that causes this.
-
Blue Cat Audio Blue Cat Audio https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=39981
- KVRAF
- 6350 posts since 8 Sep, 2004 from Paris (France)
Thanks for the details. It makes it easier to understand your problem 
.
First, the crest fator represents the ratio between the peak and average (RMS) signals. So it does not make sense for silence and is thus very sensitive to the beginning and end of your 4 bars section (which is not the case for RMS or peak values, hence the difference). So 4 bars is probably a bit short for your measurement.
Also, if you use the plug-in in real time as an insert, resetting the average value does not reset the computation of the RMS average or the peak value (some DAWs will reset the plug-in when transport is restarted but not all), so every time to restart your measurement, initial values may not be reset properly, and your measurement is thus sensitive to history.
So in order to get consistent results which actually make sense, you probably want to:
1. (optional) use a longer time period.
2. do the measurement offline using the offline processing function of your DAW (if it has one. BTW which DAW is it?).
This way the average will make sense, and the process will be repeatable because you can be sure all the plug-in meters are all properly initialized with the exact same values.
Hope this helps!
.First, the crest fator represents the ratio between the peak and average (RMS) signals. So it does not make sense for silence and is thus very sensitive to the beginning and end of your 4 bars section (which is not the case for RMS or peak values, hence the difference). So 4 bars is probably a bit short for your measurement.
Also, if you use the plug-in in real time as an insert, resetting the average value does not reset the computation of the RMS average or the peak value (some DAWs will reset the plug-in when transport is restarted but not all), so every time to restart your measurement, initial values may not be reset properly, and your measurement is thus sensitive to history.
So in order to get consistent results which actually make sense, you probably want to:
1. (optional) use a longer time period.
2. do the measurement offline using the offline processing function of your DAW (if it has one. BTW which DAW is it?).
This way the average will make sense, and the process will be repeatable because you can be sure all the plug-in meters are all properly initialized with the exact same values.
Hope this helps!