Source used by Roon for dynamic range

Love this feature. When I import music from Qobuz into Roon, this information is never present. I won’t listen to anything with compressed dynamic range and it’s a nice screening tool. If it’s conventional metadata, I’d like to urge Qobuz to deploy it.

BTW, the suggested music keeps getting better. I often have a planned listening session, only to be delightfully led away by Roon’s music suggestions. Keep up the great work!

Roon uses EBU R128 to calculate dynamic range while most people in the audio world, including the Dynamic Range database, use a crest factor algorithm. Here is something I posted on the Roon forum:

EBU R128 is perfect for volume leveling and the fact that Roon uses this to do volume leveling is great. People wanting the Dynamic Range Database DR numbers too is not a criticism of that choice. It is a realistic request based on the fact that the DR database numbers are more reflective of the actual amount of compression any given track may have. Look at this:

https://www.maat.digital/droffline/

Specifically:

“Unlike R128 and BS. 1770 meters, DR measures dynamic range from the perspective of a music engineer’s needs. In contrast, R128 and 1770 are designed to control loudness for commercials , not measure dynamic range for music, especially pop music. DR isn’t designed for broadcast loudness control, it’s purpose is to gauge the amount of dynamic range reduction, or the absence of dynamic range contrast. Designed by a member of the EBU ploud committee, the same body that created R128, DR informs an engineer about how much the mix is being or has been “stepped on,” dynamic range-wise, not about “will it pass through a broadcast chain without loudness reduction?”

I used to use volume leveling but I have turned it off. Why? Two reasons. First, it was sucking the life out of the music. It was diminishing the soundstage by making it less 3D. Second, I started listening to full albums almost exclusively so volume leveling became unnecessary as I have a remote control volume and don’t mind making an adjustment for each album. Though most of the time I don’t feel the need.

5 Likes

Excellent. Thank you. Interesting to read about the European Broadcasting Union’s 2014 standard recommendations.

Had never heard of crest factor, so thank you for new to me information. Seems straightforward. You must be an audio engineer and this is a great resource to the Roon Community.

If I’m understanding correctly the DR shown in Roon is calculated for each album/track? If true, wow. I use to screen for compressed dynamic range. Can’t stand compression/loss of detail for loudness. Interesting about volume leveling impact on soundstage.

Thank you again.

1 Like

Nope…just an audio enthusiast!

Yes, the R128 track gain, album gain, and track dynamic range are calculated. I have found the R128 dynamic range information to be worthless for me when I want to know how much a track is compressed. The crest factor method used by DR, on the other hand, provides numbers that almost always correlate to what to the actual compression of the track actually is.

You’re using DRmeter Offline to measure DR on files/batches of files?

Can it be used for streamed music or does it need the file? I’m thinking of Qobuz.

Thank you.

The DROffline tool needs a PCM file to process. It won’t work with Qobuz or Tidal.

Got it. Oliver Masciarotte (MAAT) indicates that realtime DRi measurement from streamed music requires an audiophile player that hosts plug-ins (DRMeter and DRMeter MkII). He showed me an example of output from DROffline’s formatted log. How different is this value from the Roon R128 value?

Where did you discover the method used in Roon to measure DR?

This is an interesting rabbit hole. Sounds like I’d have to be using Amarra for realtime streaming DR assessment, which is what I use to guide purchases. Not interested in going there just now.

Thank you again.

It is documented:
https://kb.roonlabs.com/Volume_Leveling
https://kb.roonlabs.com/Dynamic_Range

See also https://community.roonlabs.com/search?q=R128 for more information.

When comparing enabling and disabling volume leveling it’s important to level match the output otherwise the higher dB output will sound better. It’s how our ear/brain works.

R128 punishes those tracks/album based on how far they take the ‘loudness’ war. The more compressed, the lower the output volume.

Thank you for the pointer showing method. Have been trying to see whether Qobuz can disclose such information. Turns out the method used for measuring dynamic range isn’t as straightforward as loudness.

1 Like

Thank you. I was aware of Fletcher-Munson curves and effects, but was interested in the compression for loudness tradeoff and how this is measured. Love the fact that a form of dynamic range (DR) is in Roon and this led me to make a suggestion to Qobuz that they also provide such a measurement. Speed-Racer provided information that indicates that the R 128 method used in Roon may not be as helpful as other DR measurements. Since DR can not be measured in realtime without some additional software, I can see why Qobuz and Tidal don’t provide such useful information.

Don’t confuse Fletcher-Munson curves with the ‘loudness’ war. R128 does not change dynamic range of the music nor does it increase bass and treble at low volumes where our ears are less sensitive.

R128 is not intended nor designed to measure DR. It’s mission is to reduce the volume to a standard. Many use -16LUFS. I prefer -22LUFS but it’s all up to you in Roon to choose.

EBU R 128 is a recommendation for loudness normalisation and maximum level of audio signals. It is primarily followed during audio mixing of television and radio programmes and adopted by broadcasters to measure and control programme loudness

1 Like

Thank you. I’ve been unclear. I have been referring to the range value reported below sample rate/word length in Roon, which is a major factor in determining whether I will purchase or can listen to a recording. Music I bought previous to being able to measure this is all over the place and the recordings I enjoy are all with range > 9. The Jazz and Classical music typically runs > 12.

Sound equalization or leveling is not something I’ve actively pursued. I just dial up the volume on my DAC for this. Sounds like this is a bit tricky, given Speed_Racer’s observations. Referred to Fletcher-Munson as it refers to volume effect on sound quality and which, I assume is the point of sound equalization.

Now I’m curious about whether sound leveling is pre-applied to music I stream from Qobuz through Roon. Might explain why I hear differences between music I’ve purchased/downloaded from Qobuz versus streaming of the recording with identical sample rate/word length. I always prefer the downloaded version. Then again, may be buyer’s bias.

Thank you. You’ve got me wondering now whether streamed Qobuz music coming through Roon is being volume leveled. I like the rawest signals I can get and then turn up the volume to where I enjoy it. Speed_Racer wrote about sound stage degradation with sound leveling and that concerns me.

As from the information from the threads linked above, no (if you don’t enable volume leveling in Roon).

Well, you are probably splitting hairs. R128 does produce a value that people call “Dynamic Range” even if the standard calls it “Loudness Range”. Roon even calls it “Dynamic Range”. Whatever you call it, it’s just not very good for indicating how compressed tracks actually are.

I encourage you to give Volume Leveling a try for a week or so. I find it does a great job of evening out the experience when playing random tracks. My system is capable of significant range and I don’t like one track at X and the next at +15dB.

Continue to use your volume control as you do now.

I would caution against excluding any music simply based on it’s calculated dynamic range, irrespective of the model used. This value alone is not a terribly good indication of the quality of the recording nor reproduction.

I like to use the DR values only for comparison between different versions/masters of the same piece.

If I’m hunting the best version of a particular work I will search the DR DB - http://dr.loudness-war.info but only as a data point as the information is provided by people much like Wikipedia. People can and do mess with existing data, delete data and enter inaccurate data (either intentionally or mistakenly).

With Qobuz, Tidal and Spotify available I don’t purchase as many CDs as I used to but I will from time to time when a particular album has a strong impact on me emotionally. I always attempt to purchase directly from the artist via Bandcamp or the their web site to give them more $ and encourage what they do.

Have you heard the ‘polish’ of a well constructed convolution filter? It can be a deep hole to explore but Roon makes is quite easy to apply, once you’ve made or got the filters.

I would argue that the DR method yields incredibly good indications of how compressed recorded music actually is. You can have well recorded and well mixed music that is overly compressed.

I absolutely agree. I guess what I should have said is the level of compression is not alone a good indicator of how much the track/album will move me, cause me to tap my foot, get up and move around (dare I say dance).

I find it fatiguing to listen to overly dynamic music. The difference from the low levels to the peak can be overwhelming. Granted, it’s probably only a fraction of what I listen to that does this to me and not all the time but sometimes I do feel my system is too capable to reproducing dynamics. Rather the recordings dynamics are more than I care for at that moment in time and not limited by the reproduction.

From what I’ve read of the mastering process(es) out there, select compression is needed otherwise one instrument could drown out others.