Thoughts on music loudness and dynamic range

[jbrunwa]

Although digital media technically supports a wider dynamic range than vinyl, as I look through the following resources, it seems that many vinyl editions were produced with a wider dynamic range than digital editions, and early digital editions sometimes were produced with a wider DR than later digital editions. SACD's appear to be hit or miss with respect to DR. In general classical music and jazz seem to preserve wider DR. 

magicvinyldigital.net
dr.loudness-war.info

Or maybe it's that masters created earlier in time before compression and normalization came into vogue?

Many (especially re-issues of CD editions) seem to have compressed audio signals to get consistently loud/minimal dynamic range. Where I see comparisons of editions like at magicvinyldigital.net, not only average loudness has been increased during mastering, but audio has been clipped to reduce dynamic range.

Resources from izotope, a popular mastering tool, publish info on loudness and normalization required or performed by the popular streaming services:

www.izotope.com/en/learn/mastering-for-streaming-platforms.html
www.aes.org/technical/documentDownloads.cfm?docID=731

I guess that it would be a PITA to listen to music without normalization - the listener would have to adjust volume for each song or album. Several of the streaming services normalize both per song and per album, based on whether the user listens to a single song or more than one song from the same album.

All other things being equal, when listening to playlists' in the car or mobile/phone listening, I would prefer loudness normalized content. But I would prefer to listen to editions with the best dynamic range when I listen to complete albums at home.

I might buy some editions that were mastered with a wider DR to compare.

[marcl]

Aug 17, 2022 11:51:51 GMT -5 jbrunwa said:

Although digital media technically supports a wider dynamic range than vinyl, as I look through the following resources, it seems that many vinyl editions were produced with a wider dynamic range than digital editions, and early digital editions sometimes were produced with a wider DR than later digital editions. SACD's appear to be hit or miss with respect to DR. In general classical music and jazz seem to preserve wider DR. 

magicvinyldigital.net
dr.loudness-war.info

Or maybe it's that masters created earlier in time before compression and normalization came into vogue?

Many (especially re-issues of CD editions) seem to have compressed audio signals to get consistently loud/minimal dynamic range. Where I see comparisons of editions like at magicvinyldigital.net, not only average loudness has been increased during mastering, but audio has been clipped to reduce dynamic range.

Resources from izotope, a popular mastering tool, publish info on loudness and normalization required or performed by the popular streaming services:

www.izotope.com/en/learn/mastering-for-streaming-platforms.html
www.aes.org/technical/documentDownloads.cfm?docID=731

I guess that it would be a PITA to listen to music without normalization - the listener would have to adjust volume for each song or album. Several of the streaming services normalize both per song and per album, based on whether the user listens to a single song or more than one song from the same album.

All other things being equal, when listening to playlists' in the car or mobile/phone listening, I would prefer loudness normalized content. But I would prefer to listen to editions with the best dynamic range when I listen to complete albums. 

I might buy some editions that were mastered with a wider DR to compare.

I think you're right that as we got into the 21st century loudness wars more recent CD/download versions tend to be compressed.  I find some recordings unlistenable after being used to my many uncompressed, minimally processed recordings.

I have a version of Kind of Blue from CD, and then I got the recent remaster that was a transfer from the original 3-track source tape.  Both sound great, but the newer remaster actually measures less dynamic range than the CD.

But isn't normalization a bit different?  It should make the average consistent across tracks but should not compress the peak-to-average, right?  I use JRiver and I don't believe the Volume Leveling option affects dynamic range.

[jbrunwa]

Aug 17, 2022 12:05:45 GMT -5 marcl said:

I think you're right that as we got into the 21st century loudness wars more recent CD/download versions tend to be compressed.  I find some recordings unlistenable after being used to my many uncompressed, minimally processed recordings.

I have a version of Kind of Blue from CD, and then I got the recent remaster that was a transfer from the original 3-track source tape.  Both sound great, but the newer remaster actually measures less dynamic range than the CD.

But isn't normalization a bit different?  It should make the average consistent across tracks but should not compress the peak-to-average, right?  I use JRiver and I don't believe the Volume Leveling option affects dynamic range.

I stand corrected. Normalization does not reduce the dynamic range, while compression does. 

[KeithL]

I'm going to be a bit pedantic here and correct everyone on the terminology... (and there is a reason).

The dynamic range of all CDs is almost exactly the same....
(And it's also about 20 dB better than the physical limitation on the absolute best vinyl ever produced.)
The quietest sound on most CD discs is dead silence (technically just below -90 dB), while the loudest is within a few dB of digital 0 dB, and the dynamic range is the distance between those two limits.
Therefore, on virtually every CD, the actual dynamic range is 90-92 dB.

What has been trending lower and lower over the years is "the overall average variance in loudness".
In other words, in more modern recordings, the average loudness level tends to remain within a relatively narrow range of loudness for a higher percentage of the time.
This is important because different lists purporting to compare "dynamic range" calculate their results differently so may come up with different numbers for the same content.
(So, everyone agrees what it is in general, and measures it in a specific way, but you can't compare numbers from one list to another.)

And, to be fair, you can place most of the initial blame for this on radio.
With FM radio the actual level of the audio is independent of the strength of the FM radio carrier signal.
This means that, unlike with AM, a certain FM station will NOT be louder because it is closer, or has a more powerful transmitter.
Therefore, with FM, the only way for a song to be louder, and grab your attention as you tune across the dial, is for the music to be more heavily compressed.
And, beyond even that, with many radio listeners listening in cars, they want the quiet spots to be clearly audible over engine and wind noise.
And, in modern situations, loud enough to be audible over background noise on those cheap earbuds, or those Bluetooth speakers.
It's also a well known fact of "psychoacoustics" that, all else being equal, the system that is playing louder sounds better.
So this is one more excuse to continue this trend with streaming music and other more modern sources.

And, sadly, at this point, it has simply become a preference, or perhaps an expectation.
Some modern listeners seem to find excessive dynamic range to be jarring and to prefer a steady level...
(And, since kids these days tend to listen at a higher average level, having dynamic peaks in a recording much beyond that becomes problematic for other reasons.)
And, of course, most modern mastering software has all sort of really effective features for "maximizing loudness" - which is another term for "upwards compression and limiting".
(Raising the average level while keeping the peak levels just safely within limits.)

I'm pretty sure the main reason we no longer see "dynamic range expander" products is that modern limiters are SO effective that their results are largely irreversible...
(Although there is some modern software that does a pretty good job there too.)

Here's a link to a really excellent article on the subject in Sound On Sound Magazine:
www.soundonsound.com/sound-advice/dynamic-range-loudness-war

[marcl]

Aug 18, 2022 10:12:16 GMT -5 KeithL said:

I'm going to be a bit pedantic here and correct everyone on the terminology... (and there is a reason).

The dynamic range of all CDs is almost exactly the same....
(And it's also about 20 dB better than the physical limitation on the absolute best vinyl ever produced.)
The quietest sound on most CD discs is dead silence (technically just below -90 dB), while the loudest is within a few dB of digital 0 dB, and the dynamic range is the distance between those two limits.
Therefore, on virtually every CD, the actual dynamic range is 90-92 dB.

What has been trending lower and lower over the years is "the overall average variance in loudness".
In other words, in more modern recordings, the average loudness level tends to remain within a relatively narrow range of loudness for a higher percentage of the time.
This is important because different lists purporting to compare "dynamic range" calculate their results differently so may come up with different numbers for the same content.
(So, everyone agrees what it is in general, and measures it in a specific way, but you can't compare numbers from one list to another.)

And, to be fair, you can place most of the initial blame for this on radio.
With FM radio the actual level of the audio is independent of the strength of the FM radio carrier signal.
This means that, unlike with AM, a certain FM station will NOT be louder because it is closer, or has a more powerful transmitter.
Therefore, with FM, the only way for a song to be louder, and grab your attention as you tune across the dial, is for the music to be more heavily compressed.
And, beyond even that, with many radio listeners listening in cars, they want the quiet spots to be clearly audible over engine and wind noise.
And, in modern situations, loud enough to be audible over background noise on those cheap earbuds, or those Bluetooth speakers.
It's also a well known fact of "psychoacoustics" that, all else being equal, the system that is playing louder sounds better.
So this is one more excuse to continue this trend with streaming music and other more modern sources.

And, sadly, at this point, it has simply become a preference, or perhaps an expectation.
Some modern listeners seem to find excessive dynamic range to be jarring and to prefer a steady level...
(And, since kids these days tend to listen at a higher average level, having dynamic peaks in a recording much beyond that becomes problematic for other reasons.)
And, of course, most modern mastering software has all sort of really effective features for "maximizing loudness" - which is another term for "upwards compression and limiting".
(Raising the average level while keeping the peak levels just safely within limits.)

I'm pretty sure the main reason we no longer see "dynamic range expander" products is that modern limiters are SO effective that their results are largely irreversible...
(Although there is some modern software that does a pretty good job there too.)

Here's a link to a really excellent article on the subject in Sound On Sound Magazine:
www.soundonsound.com/sound-advice/dynamic-range-loudness-war

Excellent pedantification !     A couple things ....

I use JRiver Media Center and it analyzes dynamic range of tracks using Crest Factor DR and R128, each using a different method.  Crest Factor DR is peak to average ratio.  R128 measures how many decibels there are between a track's 95 percentile for loudness and the track's 10th percentile for loudness, essentially throwing out the high and low end of the range so very short peaks and moments of silence don't affect the measurement.

How much dynamic range can there actually be in a recording?  Mark Waldrep of AIX Records once did an orchestra recording of Revel's Bolero where he attempted to actually have close to 96db dynamic range from the quietest starting taps of the snare drum to the maximum peak of the full orchestra.  He recorded 96/24 PCM and makes a point of the "provenance" of his delivery of these recordings as 96/24 downloads (in either 2-channel or 5.1),   This recording has a Crest Factor DR of 16 and R128 of 25.1.   These are indeed very high numbers compared to many other recordings that I have ... even uncompressed live orchestras.

For comparison, the 96/24 download of Steely Dan's Gaucho album has CF-DR 7 to 11 and R128 3.9 to 8.6.   For another comparison, I recently transferred cassette recordings of a jazz quintet I played with years ago ... two horns, electronic keyboards, electric bass, large acoustic drum set ... latin jazz, fusion, etc so very wide range ... recorded with two mics on a Nakamichi 700II cassette deck wit Dolby B.  I transferred at 48KHz 16 bit.  CF-DR ranged 10-15 and R128 ranged 5.6 to 14.9.

It's interesting to listen to music and correlate how the more compressed recordings sound vs the uncompressed.  I also use an app called Music Scope to look at dynamic range and also frequency content.  This app also reveals some interesting things like how some recordings are "high res" but actual frequency response cuts off at CD quality ... and how DSD recordings have that big hump of nothing but shaped noise sucking up bandwidth.

Here's a Music Scope screen shot of a recording of Wapango by Paquito D'Rivera.  The first half is string orchestra while the second half is electric bass, piano, drums, percussion and sax.  The "polar plot" looking section shows the track playing, clockwise, with amplitude evident in the orange traces.  Clearly the strings were uncompressed and then when the jazz group comes in the dynamic range is squashed, and the average loudness goes up.  The LRA number corresponds to R128, and is 11.7 as measured by both Music Scope and JRiver.  JRiver measured the CF-DR as 14.

[Boomzilla]

Many early CDs were mastered by engineers whose mastering experience was based on vinyl records (thus the glut of “remastered” releases later). Digital mastering is different.

Aside from what’s on the CD, though, driver compression at higher volume is often responsible for the “all a loud roar” effect in many speakers.

[creimes]

Aug 18, 2022 11:35:03 GMT -5 marcl said:

Aug 18, 2022 10:12:16 GMT -5 KeithL said:

I'm going to be a bit pedantic here and correct everyone on the terminology... (and there is a reason).

The dynamic range of all CDs is almost exactly the same....
(And it's also about 20 dB better than the physical limitation on the absolute best vinyl ever produced.)
The quietest sound on most CD discs is dead silence (technically just below -90 dB), while the loudest is within a few dB of digital 0 dB, and the dynamic range is the distance between those two limits.
Therefore, on virtually every CD, the actual dynamic range is 90-92 dB.

What has been trending lower and lower over the years is "the overall average variance in loudness".
In other words, in more modern recordings, the average loudness level tends to remain within a relatively narrow range of loudness for a higher percentage of the time.
This is important because different lists purporting to compare "dynamic range" calculate their results differently so may come up with different numbers for the same content.
(So, everyone agrees what it is in general, and measures it in a specific way, but you can't compare numbers from one list to another.)

And, to be fair, you can place most of the initial blame for this on radio.
With FM radio the actual level of the audio is independent of the strength of the FM radio carrier signal.
This means that, unlike with AM, a certain FM station will NOT be louder because it is closer, or has a more powerful transmitter.
Therefore, with FM, the only way for a song to be louder, and grab your attention as you tune across the dial, is for the music to be more heavily compressed.
And, beyond even that, with many radio listeners listening in cars, they want the quiet spots to be clearly audible over engine and wind noise.
And, in modern situations, loud enough to be audible over background noise on those cheap earbuds, or those Bluetooth speakers.
It's also a well known fact of "psychoacoustics" that, all else being equal, the system that is playing louder sounds better.
So this is one more excuse to continue this trend with streaming music and other more modern sources.

And, sadly, at this point, it has simply become a preference, or perhaps an expectation.
Some modern listeners seem to find excessive dynamic range to be jarring and to prefer a steady level...
(And, since kids these days tend to listen at a higher average level, having dynamic peaks in a recording much beyond that becomes problematic for other reasons.)
And, of course, most modern mastering software has all sort of really effective features for "maximizing loudness" - which is another term for "upwards compression and limiting".
(Raising the average level while keeping the peak levels just safely within limits.)

I'm pretty sure the main reason we no longer see "dynamic range expander" products is that modern limiters are SO effective that their results are largely irreversible...
(Although there is some modern software that does a pretty good job there too.)

Here's a link to a really excellent article on the subject in Sound On Sound Magazine:
www.soundonsound.com/sound-advice/dynamic-range-loudness-war

Excellent pedantification !     A couple things ....

I use JRiver Media Center and it analyzes dynamic range of tracks using Crest Factor DR and R128, each using a different method.  Crest Factor DR is peak to average ratio.  R128 measures how many decibels there are between a track's 95 percentile for loudness and the track's 10th percentile for loudness, essentially throwing out the high and low end of the range so very short peaks and moments of silence don't affect the measurement.

How much dynamic range can there actually be in a recording?  Mark Waldrep of AIX Records once did an orchestra recording of Revel's Bolero where he attempted to actually have close to 96db dynamic range from the quietest starting taps of the snare drum to the maximum peak of the full orchestra.  He recorded 96/24 PCM and makes a point of the "provenance" of his delivery of these recordings as 96/24 downloads (in either 2-channel or 5.1),   This recording has a Crest Factor DR of 16 and R128 of 25.1.   These are indeed very high numbers compared to many other recordings that I have ... even uncompressed live orchestras.

For comparison, the 96/24 download of Steely Dan's Gaucho album has CF-DR 7 to 11 and R128 3.9 to 8.6.   For another comparison, I recently transferred cassette recordings of a jazz quintet I played with years ago ... two horns, electronic keyboards, electric bass, large acoustic drum set ... latin jazz, fusion, etc so very wide range ... recorded with two mics on a Nakamichi 700II cassette deck wit Dolby B.  I transferred at 48KHz 16 bit.  CF-DR ranged 10-15 and R128 ranged 5.6 to 14.9.

It's interesting to listen to music and correlate how the more compressed recordings sound vs the uncompressed.  I also use an app called Music Scope to look at dynamic range and also frequency content.  This app also reveals some interesting things like how some recordings are "high res" but actual frequency response cuts off at CD quality ... and how DSD recordings have that big hump of nothing but shaped noise sucking up bandwidth.

Here's a Music Scope screen shot of a recording of Wapango by Paquito D'Rivera.  The first half is string orchestra while the second half is electric bass, piano, drums, percussion and sax.  The "polar plot" looking section shows the track playing, clockwise, with amplitude evident in the orange traces.  Clearly the strings were uncompressed and then when the jazz group comes in the dynamic range is squashed, and the average loudness goes up.  The LRA number corresponds to R128, and is 11.7 as measured by both Music Scope and JRiver.  JRiver measured the CF-DR as 14.

View Attachment

I recently purchased JRiver, do you recommend using volume leveling then in the DSP settings, I currently have the box checked.

Chad

[marcl]

Aug 18, 2022 19:17:55 GMT -5 creimes said:

Aug 18, 2022 11:35:03 GMT -5 marcl said:

Excellent pedantification !     A couple things ....

I use JRiver Media Center and it analyzes dynamic range of tracks using Crest Factor DR and R128, each using a different method.  Crest Factor DR is peak to average ratio.  R128 measures how many decibels there are between a track's 95 percentile for loudness and the track's 10th percentile for loudness, essentially throwing out the high and low end of the range so very short peaks and moments of silence don't affect the measurement.

How much dynamic range can there actually be in a recording?  Mark Waldrep of AIX Records once did an orchestra recording of Revel's Bolero where he attempted to actually have close to 96db dynamic range from the quietest starting taps of the snare drum to the maximum peak of the full orchestra.  He recorded 96/24 PCM and makes a point of the "provenance" of his delivery of these recordings as 96/24 downloads (in either 2-channel or 5.1),   This recording has a Crest Factor DR of 16 and R128 of 25.1.   These are indeed very high numbers compared to many other recordings that I have ... even uncompressed live orchestras.

For comparison, the 96/24 download of Steely Dan's Gaucho album has CF-DR 7 to 11 and R128 3.9 to 8.6.   For another comparison, I recently transferred cassette recordings of a jazz quintet I played with years ago ... two horns, electronic keyboards, electric bass, large acoustic drum set ... latin jazz, fusion, etc so very wide range ... recorded with two mics on a Nakamichi 700II cassette deck wit Dolby B.  I transferred at 48KHz 16 bit.  CF-DR ranged 10-15 and R128 ranged 5.6 to 14.9.

It's interesting to listen to music and correlate how the more compressed recordings sound vs the uncompressed.  I also use an app called Music Scope to look at dynamic range and also frequency content.  This app also reveals some interesting things like how some recordings are "high res" but actual frequency response cuts off at CD quality ... and how DSD recordings have that big hump of nothing but shaped noise sucking up bandwidth.

Here's a Music Scope screen shot of a recording of Wapango by Paquito D'Rivera.  The first half is string orchestra while the second half is electric bass, piano, drums, percussion and sax.  The "polar plot" looking section shows the track playing, clockwise, with amplitude evident in the orange traces.  Clearly the strings were uncompressed and then when the jazz group comes in the dynamic range is squashed, and the average loudness goes up.  The LRA number corresponds to R128, and is 11.7 as measured by both Music Scope and JRiver.  JRiver measured the CF-DR as 14.

View Attachment

I recently purchased JRiver, do you recommend using volume leveling then in the DSP settings, I currently have the box checked.

Chad

I did for a while, then didn't, and I didn't find it that big a deal.  Maybe the kind of stuff I listen to.  But I played a LOT with the DSP Studio before I got Dirac ... did unnatural things with PEQ!

I've found so many interesting ways to use JRiver.  It's like I think up some thing I want to do, research a bit, and oh yeah, it can do that.  Very clever stuff.  And then I read something about a feature and they say "this is what this thing is meant to do but we don't think it works".

[Boomzilla]

Hi marcl -

I use(d) jRiver for Mac - and the ratio of things that are claimed to be there to those that don't work is astounding. I finally gave up and moved to Roon. Fewer options, but all the options that are available work perfectly.

[marcl]

Aug 19, 2022 12:59:04 GMT -5 Boomzilla said:

Hi marcl -

I use(d) jRiver for Mac - and the ratio of things that are claimed to be there to those that don't work is astounding. I finally gave up and moved to Roon. Fewer options, but all the options that are available work perfectly.

Interesting.  Of course they do a release at least once a year and improve things.  So far what I need has worked well.

[KeithL]

I personally use Foobar2000 for "just playing music files"  but I have used jRiver in the past.
To be fair, the parts I used always worked fine, but I didn't use most of the advanced features.
(This is the Windows version... )

However I personally DO NOT want a player program that's going to "manage my music library for me".
I prefer to store my music files in a logical structure of physical folders - which I can then play with whatever player software I like.
That way, if I want to try a different player, or switch permanently to one, I don't have to do anything to my music library itself.
(In fact, since my files aren't linked to a specific player, I can play them in Foobar2000, or jRiver, or some other player, whenever I want to.)
I can also easily find or handle specific categories...
For example, if I want to "take a copy of all of my Doors albums to work with me", I just copy the "Doors" folder with all the individual Doors album folders inside it.

In that context...
I find Foobar2000 the best...
(For example, I can right-click on a folder, and select "Enqueue in Foobar", without even opening the program first.)
And jRiver Media Center is OK... but it really wants for you to "ingest your music library" to manage it...
(You actually seem to have to go through extra steps to just have it play a separate music file without adding it to the library.)
And Roon is the most intrusive... or requires the largest commitment... however you wish to phrase it.
 

Aug 19, 2022 13:01:13 GMT -5 marcl said:

Aug 19, 2022 12:59:04 GMT -5 Boomzilla said:

Hi marcl -

I use(d) jRiver for Mac - and the ratio of things that are claimed to be there to those that don't work is astounding. I finally gave up and moved to Roon. Fewer options, but all the options that are available work perfectly.

Interesting.  Of course they do a release at least once a year and improve things.  So far what I need has worked well.

[KeithL]

MusicScope is very cool... although more "informative" than "useful".
(Unfortunately, while it's nice to be informed, as consumers we're sort of stuck with the content that is available to us.)

Sadly, while MusicScope has become a sort of standard, it looks like it is no longer available.

Aug 18, 2022 11:35:03 GMT -5 marcl said:

Aug 18, 2022 10:12:16 GMT -5 KeithL said:

I'm going to be a bit pedantic here and correct everyone on the terminology... (and there is a reason).

The dynamic range of all CDs is almost exactly the same....
(And it's also about 20 dB better than the physical limitation on the absolute best vinyl ever produced.)
The quietest sound on most CD discs is dead silence (technically just below -90 dB), while the loudest is within a few dB of digital 0 dB, and the dynamic range is the distance between those two limits.
Therefore, on virtually every CD, the actual dynamic range is 90-92 dB.

What has been trending lower and lower over the years is "the overall average variance in loudness".
In other words, in more modern recordings, the average loudness level tends to remain within a relatively narrow range of loudness for a higher percentage of the time.
This is important because different lists purporting to compare "dynamic range" calculate their results differently so may come up with different numbers for the same content.
(So, everyone agrees what it is in general, and measures it in a specific way, but you can't compare numbers from one list to another.)

And, to be fair, you can place most of the initial blame for this on radio.
With FM radio the actual level of the audio is independent of the strength of the FM radio carrier signal.
This means that, unlike with AM, a certain FM station will NOT be louder because it is closer, or has a more powerful transmitter.
Therefore, with FM, the only way for a song to be louder, and grab your attention as you tune across the dial, is for the music to be more heavily compressed.
And, beyond even that, with many radio listeners listening in cars, they want the quiet spots to be clearly audible over engine and wind noise.
And, in modern situations, loud enough to be audible over background noise on those cheap earbuds, or those Bluetooth speakers.
It's also a well known fact of "psychoacoustics" that, all else being equal, the system that is playing louder sounds better.
So this is one more excuse to continue this trend with streaming music and other more modern sources.

And, sadly, at this point, it has simply become a preference, or perhaps an expectation.
Some modern listeners seem to find excessive dynamic range to be jarring and to prefer a steady level...
(And, since kids these days tend to listen at a higher average level, having dynamic peaks in a recording much beyond that becomes problematic for other reasons.)
And, of course, most modern mastering software has all sort of really effective features for "maximizing loudness" - which is another term for "upwards compression and limiting".
(Raising the average level while keeping the peak levels just safely within limits.)

I'm pretty sure the main reason we no longer see "dynamic range expander" products is that modern limiters are SO effective that their results are largely irreversible...
(Although there is some modern software that does a pretty good job there too.)

Here's a link to a really excellent article on the subject in Sound On Sound Magazine:
www.soundonsound.com/sound-advice/dynamic-range-loudness-war

Excellent pedantification !     A couple things ....

I use JRiver Media Center and it analyzes dynamic range of tracks using Crest Factor DR and R128, each using a different method.  Crest Factor DR is peak to average ratio.  R128 measures how many decibels there are between a track's 95 percentile for loudness and the track's 10th percentile for loudness, essentially throwing out the high and low end of the range so very short peaks and moments of silence don't affect the measurement.

How much dynamic range can there actually be in a recording?  Mark Waldrep of AIX Records once did an orchestra recording of Revel's Bolero where he attempted to actually have close to 96db dynamic range from the quietest starting taps of the snare drum to the maximum peak of the full orchestra.  He recorded 96/24 PCM and makes a point of the "provenance" of his delivery of these recordings as 96/24 downloads (in either 2-channel or 5.1),   This recording has a Crest Factor DR of 16 and R128 of 25.1.   These are indeed very high numbers compared to many other recordings that I have ... even uncompressed live orchestras.

For comparison, the 96/24 download of Steely Dan's Gaucho album has CF-DR 7 to 11 and R128 3.9 to 8.6.   For another comparison, I recently transferred cassette recordings of a jazz quintet I played with years ago ... two horns, electronic keyboards, electric bass, large acoustic drum set ... latin jazz, fusion, etc so very wide range ... recorded with two mics on a Nakamichi 700II cassette deck wit Dolby B.  I transferred at 48KHz 16 bit.  CF-DR ranged 10-15 and R128 ranged 5.6 to 14.9.

It's interesting to listen to music and correlate how the more compressed recordings sound vs the uncompressed.  I also use an app called Music Scope to look at dynamic range and also frequency content.  This app also reveals some interesting things like how some recordings are "high res" but actual frequency response cuts off at CD quality ... and how DSD recordings have that big hump of nothing but shaped noise sucking up bandwidth.

Here's a Music Scope screen shot of a recording of Wapango by Paquito D'Rivera.  The first half is string orchestra while the second half is electric bass, piano, drums, percussion and sax.  The "polar plot" looking section shows the track playing, clockwise, with amplitude evident in the orange traces.  Clearly the strings were uncompressed and then when the jazz group comes in the dynamic range is squashed, and the average loudness goes up.  The LRA number corresponds to R128, and is 11.7 as measured by both Music Scope and JRiver.  JRiver measured the CF-DR as 14.

View Attachment

[Boomzilla]

Aug 19, 2022 13:01:13 GMT -5 marcl said:

Aug 19, 2022 12:59:04 GMT -5 Boomzilla said:

Hi marcl -

I use(d) jRiver for Mac - and the ratio of things that are claimed to be there to those that don't work is astounding. I finally gave up and moved to Roon. Fewer options, but all the options that are available work perfectly.

Interesting.  Of course they do a release at least once a year and improve things.  So far what I need has worked well.

My buddy uses the Windows version of jRiver & swears by it.

[marcl]

Aug 19, 2022 13:31:07 GMT -5 KeithL said:

MusicScope is very cool... although more "informative" than "useful".
(Unfortunately, while it's nice to be informed, as consumers we're sort of stuck with the content that is available to us.)

Sadly, while MusicScope has become a sort of standard, it looks like it is no longer available.

Aug 18, 2022 11:35:03 GMT -5 marcl said:

Excellent pedantification !     A couple things ....

I use JRiver Media Center and it analyzes dynamic range of tracks using Crest Factor DR and R128, each using a different method.  Crest Factor DR is peak to average ratio.  R128 measures how many decibels there are between a track's 95 percentile for loudness and the track's 10th percentile for loudness, essentially throwing out the high and low end of the range so very short peaks and moments of silence don't affect the measurement.

How much dynamic range can there actually be in a recording?  Mark Waldrep of AIX Records once did an orchestra recording of Revel's Bolero where he attempted to actually have close to 96db dynamic range from the quietest starting taps of the snare drum to the maximum peak of the full orchestra.  He recorded 96/24 PCM and makes a point of the "provenance" of his delivery of these recordings as 96/24 downloads (in either 2-channel or 5.1),   This recording has a Crest Factor DR of 16 and R128 of 25.1.   These are indeed very high numbers compared to many other recordings that I have ... even uncompressed live orchestras.

For comparison, the 96/24 download of Steely Dan's Gaucho album has CF-DR 7 to 11 and R128 3.9 to 8.6.   For another comparison, I recently transferred cassette recordings of a jazz quintet I played with years ago ... two horns, electronic keyboards, electric bass, large acoustic drum set ... latin jazz, fusion, etc so very wide range ... recorded with two mics on a Nakamichi 700II cassette deck wit Dolby B.  I transferred at 48KHz 16 bit.  CF-DR ranged 10-15 and R128 ranged 5.6 to 14.9.

It's interesting to listen to music and correlate how the more compressed recordings sound vs the uncompressed.  I also use an app called Music Scope to look at dynamic range and also frequency content.  This app also reveals some interesting things like how some recordings are "high res" but actual frequency response cuts off at CD quality ... and how DSD recordings have that big hump of nothing but shaped noise sucking up bandwidth.

Here's a Music Scope screen shot of a recording of Wapango by Paquito D'Rivera.  The first half is string orchestra while the second half is electric bass, piano, drums, percussion and sax.  The "polar plot" looking section shows the track playing, clockwise, with amplitude evident in the orange traces.  Clearly the strings were uncompressed and then when the jazz group comes in the dynamic range is squashed, and the average loudness goes up.  The LRA number corresponds to R128, and is 11.7 as measured by both Music Scope and JRiver.  JRiver measured the CF-DR as 14.

View Attachment

Yes, unfortunately.  I looked around and one site supposedly had a download but I couldn't find it.  I gave someone my account info to activate but their activation server is gone too.  Fortunately I activated mine a couple years ago so I'm good until I get a new PC.

I like checking actual music bandwidth, the dynamic range plot of course, and also the stereo separation plot ... some music having surprisingly little separation.

[MusicHead]

Perhaps this?

musicscope.software.informer.com/download/#downloading