I was extremely intrigued by the latest Erin’s Audio Corner video on YT about Dynamic Range and it got me really thinking about what I’d recently been buying online from Qobuz compared with what I’d been buying ages ago in CD format.
I was trying to think of an early CD I’d bought and see what the DR was and remembered Depeche Mode’s debut album, Speak & Spell. So, off I stomped up into the attic and luckily/weirdly found what I was looking for in the first box I came across. Specifically it’s this CD here with the exact matching barcode.
Now on this website it claims the DR is 14 (assumed average); minimum track is 11 and maximum track is 16.
I ripped the CD using Apple Music in AIFF format. I ran MAAT DROffline MKII on the CD and the results were very similar to the website above.
Roon however is saying the DR is 6 (which according to the DR database website is rather naff).
I repeated the rip again, this time using WAV. Same results.
Does anyone know why this is or how DR is calculated in Roon? I’m not sure whether this is something I am misunderstanding/doing wrong or whether it’s a bug in Roon (or even something else). I was wondering whether anyone else has found any disparity between the DR database and Roon?
Thanks for writing in - and excellent question you have here! Sounds like a fun process as well (nice work finding the specific album you were looking for as well, I know how lucky that can feel!)
Roon uses its own algorithm, based on the EBU R128 standard, which is different from the older TT DR Meter (used by the Dynamic Range Database and MAAT DROffline MKII). The two methods aren’t directly comparable - they measure different aspects of the signal and use different statistical weighting.
You’re not doing anything wrong! Your rips are valid, and your MAAT results are correct within the TT DR framework. Roon’s DR measurement is not a bug, but a different standard, useful for relative comparison within Roon, but not interchangeable with DR Database figures.
I would take from this (again), that DR should be completely ignored. It says basically nothing about the recording, not to speak of sound or mastering quality.
@benjamin thank you for spelling this out in the response I hadn’t appreciated there were two calculation methods.
What I was hoping to do after watching the video was sniff out all the bad recordings I have and possibly replace them with better ones.
I was listening to some music I really enjoy the other day but kept squirming at how bad the recording was - it was just like Erin had described and really hit home.
What I’m trying to understand right now then is how the current Roon algorithm can help me achieve my goal or whether the Crest algothirm is better suited to this? If the latter then I need to go and upvote that feature request!
I’m not even sure how to do this. Without even trying I would expect them to be what Roon says they are based on the feedback so far. I hadn’t appreciated there were two calculation methods and Roon uses the other one. What I am trying to do is sniff out the bad recordings and then possibly source better ones. I think it’s terribly silly that I’ve possibly paid for a high-res recording thinking I was getting a better version than a low-res one I already owned. But hey ho you live and learn!
Aye. Just be aware that this is a huge rabbit hole - you will have to decide if you actually enjoy the research needed to find that elusive ‘best’ master, or whether you’d rather just listen to music. Either is valid - it’s your choice…
And don’t believe everything you read on t’interweb.
I enjoy searching out those high(er) DR releases, for me it’s part of the hobby… which would be made much easier if Roon displayed “crest factor”. Some people are happy with Spotify, and that’s ok too.
That it exactly what comparing DR cannot give. DR is a mathematical calculation, and vastly dependent on the instrumentation and dynamic of how musicians play over the course of a track or album.
I have albums which are regarded to be most wonderfully sounding, done by the most audiophile record labels in the world, showing DR3, DR4 or DR5 simply because of the instrumentation. It is completely useless to judge sound quality like this.
Neither algorithm can help you with that. Listen to the recordings and judge yourself!
You would not know before you listen yourself.
A lower DR of different mastering versions of one and the same recording could have a bunch of possible explanations. Some mastering engineer might have adjusted the level normalization between tracks manually, clicks or inaudible subsonic noises might have been removed, peak limiting applied, bandwidth-limited compression applied or whatever. All these measures drastically change the DR but it do not necessarily mean the sound quality has worsened.
I guess the OP was talking about selecting music by sound quality, not by selecting higher DR figures. It is two issues not related to each other.
