Music services that display dynamic range/loudness?

[Eetu]

I've noticed that many of my early 90s CDs are better than the versions available currently on streaming services.

To use a couple of well-known albums as an example flacs ripped from my CDs of Nirvana 'Nevermind' and Nas 'Illmatic' both have an average DR of ~11 whereas the best versions currently available for streaming are apparently something like ~6.

And similarly, if an album has several versions available for streaming the level of compression can vary a lot between the versions.

That got me wondering, which music players/servers/streaming services display either dynamic range and/or loudness (R 128)? And are there any that show it on a track-by-track basis for easy comparison?

I have a large collection of flacs with DR and ReplayGain tags tagged with foobar2000. But anything I don't have to use a PC with? Or where I don't have to click several times to read the tags would be nice. Do I really have to get Roon or are there any alternatives?

Let me know what your solutions are, thanks.

ps. currently mainly using a Raspberry Pi running Plexamp and Logitech Media Server and also stream via Spotify.

pps. I know that DR and loudness don't guarantee the best mastering and don't tell us much about spectral content but let us focus on the aforementioned for the sake of this thread.

 

[OMas]

Measuring CD rips is easy, but the only way I know to measure DRi through a streaming app is to use the native version of DRMeter MkII. As one of the “DR guys,” I gotta warn you that foobar’s so called DR measurement isn’t. We at MAAT Labs have run comparison tests, and their algorithm is inaccurate. Not sure, but it seems to actually be a simple crest factor measurement. Same goes for BX plug–ins and other DR clones.

There are only three ways to measure actual DRi or integrated DR as listed in the Dynamic Range DB. They are:

• DROffline
• DRMeter MkII
• DROffline MkII

Our DRMeter plug–in can estimate DRi but cannot measure it.

P.S. - All the vinyl rips in the Dynamic Range DB can be ignored. You cannot accurately measure DRi from vinyl or from a lossy–compressed file or stream. So, unless you’re using DRMeter MkII and streaming lossless data with Loudness compensation disabled, you’ll always get inaccurate results.

 

[radix]

Roon has an option to display dynamic range (R128 loudness) for your library.

IIRC, I used MusicBrainz to import my CD's with the ReplayGain plugin to add the right loudness tags. Roon can then use those tags when it imports the file.

 

[Elkerton]

I wonder if the following website might be of use, though I don't know the method used:

Album list - Dynamic Range DB

 

[OMas]

Roon has an option to display dynamic range (R128 loudness) for your library.

IIRC, I used MusicBrainz to import my CD's with the ReplayGain plugin to add the right loudness tags. Roon can then use those tags when it imports the file.

LRA or the generic “Loudness” metric in R128 was designed for automated (machine) compensation of subjective loudness for all video program types (adverts, movies, trailers, serials, etc.), not music and especially not popular music. Music genres with extreme dynamic, like classical, folk and some jazz, are very difficult to measure in any meaningful way. While the 1770 family of measurements is repeatable almost everywhere in the world now, there are exceptional countries where the measurements are skewed, LRA doesn’t “translate” very well into a human’s perception of “how loud.” Works great for broadcast automation though!

As to ReplayGain; it’s not standardized though, in general, it is widely adopted across the Windows universe and the spec is freely available. The resulting metric does not bear any relationship to 1770 Loudness nor DRi.

 

[Eetu]

Thanks @OMas for clarification. I'll definitely give DROffline MkII a try.

I take it many of MAAT Labs customers use the products with DAWs. But my interest is primarily my digital music library (and streaming). Do you happen to have any music player recommendations that display the metadata created by, for example, DRO? As mentioned I'm interested if there's something available that would allow you to easily read, compare or even sort by DR(i).

 

[OMas]

Correct; with DAWs. There are a select few players that “host” plug–ins. Both my two reference players do. They are Audirvana and Amarra Luxe. For Windows, you would use a VST 2 or VST 3 plug–in, and for macOS, you’d use the AU format.

As I mentioned above, our entry level DRMeter plug–in can estimate DRi but DRMeter MkII can measure that metric in real time. DRi, like LRA, is an “integrated” or integrated–over–time measurement, that’s what the i stands for, where the entire song is measured to derive the value. DRMeter MkII can perform that measurement as you play a song.

DRMeter MkII also ships with a “native” version that’s not a plug–in. It runs standalone, as an application. Using a virtual driver, you can “wire” its input from any player and stream audio through DRM2 in real time. The product page and free–to–download user manual discuss that.

 

[little-endian]

I've noticed that many of my early 90s CDs are better than the versions available currently on streaming services.

Hmm, why am I not surprised. Would also be way too easy to just take over the decent masters from back in the days. No, of course, some moron has to mangle it around again and again.

pps. I know that DR and loudness don't guarantee the best mastering

True, but if some simple number value is all you have for comparison, it is an astonishingly good indicator. Or rather the reverse can be proven - if it is in the one-digit-range, it cannot be a great mastering anymore as virtually any content would be even better if more dynamic.

P.S. - All the vinyl rips in the Dynamic Range DB can be ignored. You cannot accurately measure DRi from vinyl

Yes, like Ian Shepherd also once noticed (where as a side topic I already humbly diagree. Certainly not everyone agrees that this album sounds great, but some including me rather think "well, given how idiotically compressed this thing is as usual, it admittedly still sounds relatively decent, but could be even so much better with DR > 14.")

Anyway, has this antique-vinyl nonsense vs. PCM discrepancy meanwhile been sorted out? Of course, one has tops and crackles, but do they impact the nominal measured dynamic range to such extend?

or from a lossy–compressed file or stream.

Which is even more bizarre as typical common lossy codecs again all audiophile non-sense claims don't reduce the dynamic range. So if tools like DR are mislead, it should be due to the introduced (mostly imperceivable) artifacts of the psycho-acoustic models.

 

[OMas]

It has nothing to do with dynamic range per se…For vinyl, it’s a simple fact that every playback chain is uncalibrated (the phono section and preamp) and includes a highly variable physical transducer; the arm and cartridge. Hence, no way to accurately measure anything except a stand-alone measurement of one person’s rig at one point in time.

In the case of lossy codecs, it’s simply that encoders are non–standardized (only the decoders are spec’d) and, being lossy, there are unknown aspects of the original data that has been removed. Hence, any measurement cannot “know” what the original input data stream was. See <https://maat-digital-downloads.s3.us-west-1.amazonaws.com/public/docs/DRMeter_UM.pdf>, page 13 - “PCM vs MP3 DR, ” for more in depth info.

With an optical disc; assuming no error concealment, then the data stream is deterministic (completely repeatable) from one player/transport to another and one disc to another.

 

[DVDdoug]

Of course you can't fully analyze a stream until after you've received the whole stream.

All of the popular streaming services use loudness normalization so all streams from the same source should be about equal in loudness. They don't (normally) use any dynamic compression so the dynamics are unaffected. But it's had a (small) indirect effect on dynamics with newer productions because producers & engineers can't just compress, limit, and boost to win the loudness war.

The exception is highly-dynamic quiet-sounding music where the target loudness can't be reached without clipping the peaks. But most tracks are reduced in volume so there is still less difference than before the loudness normalization/matching. ReplayGain works the same way (but with it's own unique algorithm)... The default is not to allow clipping so the boost is limited on some songs but it can be optionally allowed if you want to go-ahead and boost those quiet songs into clipping.

Lossy file compression (MP3, etc.) tends to increase the crest factor ("DR") without affecting the sound of the dynamics. Some new peaks are higher and some are lower but those short-term peaks aren't heard as "loudness" peaks Something similar happens with the vinyl cutting & playback process. So the MP3 and vinyl will usually "measure better" than the CD, even if they are made from the same master. Usually that's just a couple of dB or less. Theoretically it should have a bigger effect on highly (dynamically) compressed/limited recordings. Most of (or at least half) the MP3s I've ripped from CDs go over 0dB when opened in Audacity. (And MP3 can actually go over 0dB without clipping.)

one has tops and crackles, but do they impact the nominal measured dynamic range to such extend?

The pops & crackles shouldn't make too much difference unless you have a big click louder than anything else. It's mostly related to the all-pass filtering effects of the RIAA record & playback EQ, which shifts the phase of different frequencies differently and changes the wave shape. (That makes the peaks at different frequencies line-up differently.) The mechanical cutting & playback may also introduce these kinds of phase-shifts.

But, we almost never know if the vinyl and CD used the same master unless you have a record from the analog days and the CD says "remastered". There were "loudness wars" in the vinyl days but they didn't have the advanced "digital weapons".

 

[danadam]

Spotify provides 'loudness' through its API but I'm not sure how exactly it was measured.

Web API Reference | Spotify for Developers

developer.spotify.com

 

[benanders]

I've noticed that many of my early 90s CDs are better than the versions available currently on streaming services.

…

That got me wondering, which music players/servers/streaming services display either dynamic range and/or loudness (R 128)? And are there any that show it on a track-by-track basis for easy comparison?

I have a large collection of flacs with DR and ReplayGain tags tagged with foobar2000. But anything I don't have to use a PC with? Or where I don't have to click several times to read the tags would be nice.

Why is displaying DR from streaming services (I assume you mean on a mobile device / mobile app) a priority for you?
Is it to facilitate identification of any disparity between the CD copy vs. streamed copy for a given track / album?

 

[Eetu]

Why is displaying DR from streaming services (I assume you mean on a mobile device / mobile app) a priority for you?
Is it to facilitate identification of any disparity between the CD copy vs. streamed copy for a given track / album?

Yes, for example. And since many older albums may have multiple versions available it would be easy to find the remasters least affected by the loudness wars.

While not top priority it would certainly be a value-adding feature.

In the long run, I would like to own the best versions of all my favorite albums. If I would only rely on streaming I would be stuck with inferior versions ('Illmatic' and 'Nevermind' as mentioned in the op) of many albums as the OG versions aren't available, only 'loud' remasters.

I also own roughly 500 various artists compilations, many not available for streaming due to licensing etc. Having the track DR info easily available would help to identify which has the cleanest version of a certain classic disco song. Or which John Coltrane box set has the best dynamics and so on.

 

[goat76]

Yes, for example. And since many older albums may have multiple versions available it would be easy to find the remasters least affected by the loudness wars.

While not top priority it would certainly be a value-adding feature.

In the long run, I would like to own the best versions of all my favorite albums. If I would only rely on streaming I would be stuck with inferior versions ('Illmatic' and 'Nevermind' as mentioned in the op) of many albums as the OG versions aren't available, only 'loud' remasters.

I also own roughly 500 various artists compilations, many not available for streaming due to licensing etc. Having the track DR info easily available would help to identify which has the cleanest version of a certain classic disco song. Or which John Coltrane box set has the best dynamics and so on.

As you said yourself in the end of your first post, the DR number can't really be used as a quality check for which master is the “best” one, at least not as long as it’s just a few dB’s difference. Only when the difference is large as in your example with the Nevermind album, it’s a clear indication of unnecessary reduction of the dynamics.

If a certain master has say 2 dB less dynamic range than another, it's still possible this master sounds more rightly balanced tonality-wise than the original master. It may be that the originally released master was thin sounding and this problem was addressed in the remastered version, but more bass will almost always decrease the dynamic range in the measurement.

So what to do?
We have to listen and then decide which one of the masters sounds the best and most balanced, at least between the masters that are not obviously dynamically limited like your example with the Nevermind album.

 

[dasdoing]

only one that comes to my mind is Youtube. in "stats for nerds" you can see how much it was turned down to match -14LUFS, or how much it is under that if that is the case

 

[benanders]

Yes, for example. And since many older albums may have multiple versions available it would be easy to find the remasters least affected by the loudness wars.

Your reasoning seems fair enough, given the recognition that an album’s DR (a very specific measurement) isn’t necessarily pinpoint as an overall indicator of likability (a broadly subjective rank-order at best).
Not necessarily easy to make happen though: DR provided for streamed albums might work to your advantage only if (1) at least one of the bigger streaming services adopted the practice and (2) all of their tracks / albums were “acquired” from the original sources (digitized production masters, algorithm-converted/formatted files, cd rips, etc./whatever…) and evaluated for DR using the same process.
Keeping everything balanced would be a real undertaking that’d offer minimal commercial benefit and probable drawbacks - what if your streaming service had low DR measurements and you were the only service showing that!? Without knowing the true album origins (that #2 requirement above, also often lacking from streaming services yet even more important than, say, DR as a testament to subjective “quality” ), the DR metric for any given track might not be true-to-source, anyway. Available DR measurements would apply to the streamed file you hear, but would not necessarily hold true to the given master/remaster from which it was acquired. Does that make sense? Because streamed files are a product at the end of variable, unique and often independent processing x packaging strategies.

While not top priority it would certainly be a value-adding feature.

In the long run, I would like to own the best versions of all my favorite albums. If I would only rely on streaming I would be stuck with inferior versions ('Illmatic' and 'Nevermind' as mentioned in the op) of many albums as the OG versions aren't available, only 'loud' remasters. …

These are examples of albums heavily processed from the get-go. Without knowing details of the production chain, I suspect the most rewarding course of action would be to listen to as many versions from originally circulated sources as is convenient for you, given your goal is anchored in subjectivity.
To want the “best” is dependent on a number of things all specific to you. Given you’ll have a hard time demonstrating that DR is even a reliable indicator of what you deem best overall in a track/album, you could well be putting the cart before the horse.
Worth considering:
Knowing DR for a given version of a track/album before/during your listening evaluation will introduce unavoidable bias, especially since you’ve already drawn the assumption higher DR = higher likelihood of better overall sound. To find your best version of a track/album, you’d need to (1) listen to all available versions (some independence / repeated measures issues there, but…), (2) rank-order the versions based on how they sound, then (3) see if rank-orders correlate with DR figures (or any other measurements). Going the other way, the way you presently want to, might be working against instincts that would better serve your purpose.

As you said yourself in the end of your first post, the DR number can't really be used as a quality check for which master is the “best” one, at least not as long as it’s just a few dB’s difference. Only when the difference is large as in your example with the Nevermind album, it’s a clear indication of unnecessary reduction of the dynamics.

…

So what to do?
We have to listen and then decide which one of the masters sounds the best and most balanced, at least between the masters that are not obviously dynamically limited like your example with the Nevermind album.

I agree with @goat76 recommendation. Especially given @Eetu wants to own these tracks / albums and not rely on indefinite streaming subscription(s) (am I correct? ), it will give superior results to abiding arbitrary numerical indicators for the chosen recordings.

My own similar interest is exactly why I (1) do not use music streaming subscriptions and (2) use a very bare-bones music transport app.
I keep multiple versions of many albums, and the more lavish app UI’s rely too heavily on limited, conflicting or unavailable metadata that complicate keeping > 1 master / press / release of an album on hand for quick and easy playback.
Identifying one or more keeper tracks / albums for a large personal digital music library is time-consuming, but then again so is subsequently enjoying those “best” selections; that we should all be so lucky, eh!?

 

[ppataki]

Measuring CD rips is easy, but the only way I know to measure DRi through a streaming app is to use the native version of DRMeter MkII. As one of the “DR guys,” I gotta warn you that foobar’s so called DR measurement isn’t. We at MAAT Labs have run comparison tests, and their algorithm is inaccurate. Not sure, but it seems to actually be a simple crest factor measurement. Same goes for BX plug–ins and other DR clones.

There are only three ways to measure actual DRi or integrated DR as listed in the Dynamic Range DB. They are:

• DROffline
• DRMeter MkII
• DROffline MkII

Our DRMeter plug–in can estimate DRi but cannot measure it.

P.S. - All the vinyl rips in the Dynamic Range DB can be ignored. You cannot accurately measure DRi from vinyl or from a lossy–compressed file or stream. So, unless you’re using DRMeter MkII and streaming lossless data with Loudness compensation disabled, you’ll always get inaccurate results.

I have tried both the offline and the real-time versions and they are both awesome! Great job!