theSC00BZ
New Head-Fier
- Joined
- Nov 10, 2010
- Posts
- 26
- Likes
- 15
Hey All,
Let me start off by saying that I'm categorically *not* an expert or have any professional experience in audio engineering. I'm just a hippy, home, audiophile with an appreciation of good sounding music. I'm running most of my stuff from JRiver & FLAC > Ciúnas DAC > MF M1-HPA > Grado GS1000i (or Denon AH-D7000). I also have a QSP Elite and some Monitor Audio speakers for open listening, but that's for background stuff; my apartment wasn't really built with acoustics in mind. (I'm certain the upstairs neighbour walking around on a wooden floor wasn't taken into consideration.) I also recently got an iBasso DX90 for portable stuff... but I'm not here to write about equipment, only to provide some background. Again I'm not an expert, but I love music.
Up until recently, I hadn't given much thought on the technical side of sound engineering - specifically CD mastering - so I figured I'd try to understand some stuff and educate myself a little. I read articles all about the "loudness wars" and further started to understand more about dynamic range in audio. So arming myself with the TT Dynamic Range Meter I set to work.
As it happened, a while ago I bought two versions of Yazoo's: Upstairs at Eric's CDs. The original, and a remastered version. Don't roll your eyes. It was an episode of Fringe when Walter was wandering around outside in his robe, sat in an abandoned car, found a mix CD and put it on. Amazingly it still worked and, Yazoo's 'Only You' started to play. It was a touching scene. Bite me.
Anyways... I didn't know whether to get the original or remastered, and as they were relatively cheap, I got both and figured I'd use this as an opportunity to compare an original and remastered version. Well, long story short, they went through my usual ripping process and were added to my library, but I never got around to critically comparing them. I made a quick check when I first ripped them and could immediately hear that the remastered version was much louder, but that's as far as I went at the time.
Now this brings me to last night. I decided to "try" and critically compare the two, to see which one I preferred and which one I could delete from my library. After applying replay gain tags (to balance the levels) and ABX'ing them in foobar, they sounded quite similar (from my lamen perspective) - however one version seemed to have a minuscule reverb/echo to all the music. I'm sure that's not the technical term or reason for it, but I know I didn't like it. It wasn't random either, I could ABX it all the time. Turns out it was the remastered version. Based on this alone I immediately preferred the original.
After that, I ran them through the TT Dynamic Range Meter with the following results:
----------------------------------------------------------------------------------------------
Yazoo - Upstairs at Eric's (Original)
----------------------------------------------------------------------------------------------
DR Peak RMS Filename
----------------------------------------------------------------------------------------------
DR15 -1.26 dB -18.73 dB 0101 - Don't Go.wav
DR15 -1.37 dB -20.21 dB 0102 - Too Pieces.wav
DR16 -2.51 dB -21.03 dB 0103 - Bad Connection.wav
DR16 -3.70 dB -24.40 dB 0104 - I Before E Except After C.wav
DR17 -3.18 dB -23.84 dB 0105 - Midnight.wav
DR18 -1.23 dB -22.45 dB 0106 - In My Room.wav
DR14 -2.80 dB -20.32 dB 0107 - Only You.wav
DR16 -1.44 dB -20.52 dB 0108 - Goodbye 70's.wav
DR16 -2.56 dB -21.25 dB 0109 - Tuesday.wav
DR15 -4.74 dB -23.74 dB 0110 - Winter Kills.wav
DR17 -1.22 dB -19.39 dB 0111 - Bring Your Love Down (Didn't I).wav
----------------------------------------------------------------------------------------------
Official DR value: DR16
==============================================================================================
----------------------------------------------------------------------------------------------
Yazoo - Upstairs at Eric's (Remastered)
----------------------------------------------------------------------------------------------
DR Peak RMS Filename
----------------------------------------------------------------------------------------------
DR11 -0.10 dB -12.57 dB 0101 - Don't Go.wav
DR11 -0.10 dB -14.14 dB 0102 - Too Pieces.wav
DR11 -0.10 dB -12.76 dB 0103 - Bad Connection.wav
DR12 -0.10 dB -16.17 dB 0104 - I Before E Except After C.wav
DR11 -0.10 dB -13.46 dB 0105 - Midnight.wav
DR11 -0.10 dB -13.85 dB 0106 - In My Room.wav
DR10 -0.10 dB -12.02 dB 0107 - Only You.wav
DR10 -0.10 dB -11.33 dB 0108 - Goodbye 70's.wav
DR11 -0.10 dB -12.47 dB 0109 - Tuesday.wav
DR13 -0.10 dB -16.36 dB 0110 - Winter Kills.wav
DR11 -0.10 dB -11.68 dB 0111 - Bring Your Love Down (Didn't I).wav
----------------------------------------------------------------------------------------------
Official DR value: DR11
==============================================================================================
Blasphemy! Now I'm sure there are much worse examples out there, but seeing this, I couldn't believe the difference. Why would anyone do such a thing? Below is the waveform of track#5: Midnight (original and remastered converted to mono for a smaller display).
I'm assuming this is "loudness wars" hard at work.
The problem is, for the average person - myself included - it takes time and effort to understand it. Consequently, not many people bother. I had an idea. I dabble a little in photography (again not professionally), but I understand a little more about the technical aspects. Photography also has a dynamic range, and I wondered if I could try and represent the dynamic range compression in audio, in a visual format that might be a little easier to understand. It should be possible to represent the change happening in audio, in a similar format to pictures, so I Googled for a couple of pics, opened up Excel to make some calculations and Photoshop for the adjustments.
Citing wiki: "Observed 16-bit digital audio dynamic range is about 90 dB." (Yeah I know, in theory it's 96dB, and I'm sure there are other things also, but this Lamen needs to keep it simple.) Applying that crudely to RGB in photography - we can work with 255 as an equivalent. Thus, my current hair-brained, Wile E. Coyote master plan has a scale. 0dB = 255 (all channels) and -90dB = 0 (all channels). I have my light and dark values.
Using track#5 Midnight as the example, let's try applying the relative differences between the original and remastered version to a couple of pictures. First the numbers...
From a peak of -3.18 dB to a peak of -0.10 dB on a 90dB scale, we have a relative increase of 3.42%
From a RMS of -23.84 dB to an excruciating RMS of -13.46 dB on a 90dB scale, we have a relative increase of 11.53%
Using those percentages on the RGB scale, let's go into Photoshop. Here's my first example:
Using the historgram (that funky looking chart towards the top-right) - there is a 'Mean' RGB value underneath it of 43.08. I want to increase that by 3.42% (0-255). To do this, I'm going to increase the exposure to approximate the increased volume (- lamen terms remember).
This take us to 50.98. Now it doesn't look that bad at the moment. All we've actually done is made it brighter (or louder). However, now we need to reduce the range. I'm going to do this with the 'Levels'. We need to increase (or technically decrease) it by 11.5% The Levels slider that I'm using has a max of 10 and starts with 1, so I'm going to increase this to 2.15
This has (in a basic way) decreased the dynamic range of the image. In this example, you've lost some of the darker (quieter) detail in the shadows and increased their brightness (volume).
Here's the before and after:
This is more of a subtle difference, and you can make arguments for the changes depending on your preference. However, by making all the shadows brighter, you've lost huge contrast between the bright ray of light and the dark shadows.
Now what happens if we apply the same logic to another picture:
Let's start again by increasing the exposure
Then let's increase the levels
And finally the before and after.
I'm sure most people would agree that the original picture is better. There is an intention to have dark and light areas. It's the same with music, there's an intention to have loud moments and quiet moments. When you compress the dynamic range however, you have less and less space to separate the bright from the dark, the loud from the quiet.
Now feel free to rip me a new one or update/educate me in my comparisons
Pic source:
http://www.superbwallpapers.com/nature/ray-of-sun-through-the-canyon-36249/
http://www.superbwallpapers.com/nature/starry-night-sky-over-the-mountains-14678/
Let me start off by saying that I'm categorically *not* an expert or have any professional experience in audio engineering. I'm just a hippy, home, audiophile with an appreciation of good sounding music. I'm running most of my stuff from JRiver & FLAC > Ciúnas DAC > MF M1-HPA > Grado GS1000i (or Denon AH-D7000). I also have a QSP Elite and some Monitor Audio speakers for open listening, but that's for background stuff; my apartment wasn't really built with acoustics in mind. (I'm certain the upstairs neighbour walking around on a wooden floor wasn't taken into consideration.) I also recently got an iBasso DX90 for portable stuff... but I'm not here to write about equipment, only to provide some background. Again I'm not an expert, but I love music.
Up until recently, I hadn't given much thought on the technical side of sound engineering - specifically CD mastering - so I figured I'd try to understand some stuff and educate myself a little. I read articles all about the "loudness wars" and further started to understand more about dynamic range in audio. So arming myself with the TT Dynamic Range Meter I set to work.
As it happened, a while ago I bought two versions of Yazoo's: Upstairs at Eric's CDs. The original, and a remastered version. Don't roll your eyes. It was an episode of Fringe when Walter was wandering around outside in his robe, sat in an abandoned car, found a mix CD and put it on. Amazingly it still worked and, Yazoo's 'Only You' started to play. It was a touching scene. Bite me.
Anyways... I didn't know whether to get the original or remastered, and as they were relatively cheap, I got both and figured I'd use this as an opportunity to compare an original and remastered version. Well, long story short, they went through my usual ripping process and were added to my library, but I never got around to critically comparing them. I made a quick check when I first ripped them and could immediately hear that the remastered version was much louder, but that's as far as I went at the time.
Now this brings me to last night. I decided to "try" and critically compare the two, to see which one I preferred and which one I could delete from my library. After applying replay gain tags (to balance the levels) and ABX'ing them in foobar, they sounded quite similar (from my lamen perspective) - however one version seemed to have a minuscule reverb/echo to all the music. I'm sure that's not the technical term or reason for it, but I know I didn't like it. It wasn't random either, I could ABX it all the time. Turns out it was the remastered version. Based on this alone I immediately preferred the original.
After that, I ran them through the TT Dynamic Range Meter with the following results:
----------------------------------------------------------------------------------------------
Yazoo - Upstairs at Eric's (Original)
----------------------------------------------------------------------------------------------
DR Peak RMS Filename
----------------------------------------------------------------------------------------------
DR15 -1.26 dB -18.73 dB 0101 - Don't Go.wav
DR15 -1.37 dB -20.21 dB 0102 - Too Pieces.wav
DR16 -2.51 dB -21.03 dB 0103 - Bad Connection.wav
DR16 -3.70 dB -24.40 dB 0104 - I Before E Except After C.wav
DR17 -3.18 dB -23.84 dB 0105 - Midnight.wav
DR18 -1.23 dB -22.45 dB 0106 - In My Room.wav
DR14 -2.80 dB -20.32 dB 0107 - Only You.wav
DR16 -1.44 dB -20.52 dB 0108 - Goodbye 70's.wav
DR16 -2.56 dB -21.25 dB 0109 - Tuesday.wav
DR15 -4.74 dB -23.74 dB 0110 - Winter Kills.wav
DR17 -1.22 dB -19.39 dB 0111 - Bring Your Love Down (Didn't I).wav
----------------------------------------------------------------------------------------------
Official DR value: DR16
==============================================================================================
----------------------------------------------------------------------------------------------
Yazoo - Upstairs at Eric's (Remastered)
----------------------------------------------------------------------------------------------
DR Peak RMS Filename
----------------------------------------------------------------------------------------------
DR11 -0.10 dB -12.57 dB 0101 - Don't Go.wav
DR11 -0.10 dB -14.14 dB 0102 - Too Pieces.wav
DR11 -0.10 dB -12.76 dB 0103 - Bad Connection.wav
DR12 -0.10 dB -16.17 dB 0104 - I Before E Except After C.wav
DR11 -0.10 dB -13.46 dB 0105 - Midnight.wav
DR11 -0.10 dB -13.85 dB 0106 - In My Room.wav
DR10 -0.10 dB -12.02 dB 0107 - Only You.wav
DR10 -0.10 dB -11.33 dB 0108 - Goodbye 70's.wav
DR11 -0.10 dB -12.47 dB 0109 - Tuesday.wav
DR13 -0.10 dB -16.36 dB 0110 - Winter Kills.wav
DR11 -0.10 dB -11.68 dB 0111 - Bring Your Love Down (Didn't I).wav
----------------------------------------------------------------------------------------------
Official DR value: DR11
==============================================================================================
Blasphemy! Now I'm sure there are much worse examples out there, but seeing this, I couldn't believe the difference. Why would anyone do such a thing? Below is the waveform of track#5: Midnight (original and remastered converted to mono for a smaller display).
I'm assuming this is "loudness wars" hard at work.
The problem is, for the average person - myself included - it takes time and effort to understand it. Consequently, not many people bother. I had an idea. I dabble a little in photography (again not professionally), but I understand a little more about the technical aspects. Photography also has a dynamic range, and I wondered if I could try and represent the dynamic range compression in audio, in a visual format that might be a little easier to understand. It should be possible to represent the change happening in audio, in a similar format to pictures, so I Googled for a couple of pics, opened up Excel to make some calculations and Photoshop for the adjustments.
Citing wiki: "Observed 16-bit digital audio dynamic range is about 90 dB." (Yeah I know, in theory it's 96dB, and I'm sure there are other things also, but this Lamen needs to keep it simple.) Applying that crudely to RGB in photography - we can work with 255 as an equivalent. Thus, my current hair-brained, Wile E. Coyote master plan has a scale. 0dB = 255 (all channels) and -90dB = 0 (all channels). I have my light and dark values.
Using track#5 Midnight as the example, let's try applying the relative differences between the original and remastered version to a couple of pictures. First the numbers...
From a peak of -3.18 dB to a peak of -0.10 dB on a 90dB scale, we have a relative increase of 3.42%
From a RMS of -23.84 dB to an excruciating RMS of -13.46 dB on a 90dB scale, we have a relative increase of 11.53%
Using those percentages on the RGB scale, let's go into Photoshop. Here's my first example:
Using the historgram (that funky looking chart towards the top-right) - there is a 'Mean' RGB value underneath it of 43.08. I want to increase that by 3.42% (0-255). To do this, I'm going to increase the exposure to approximate the increased volume (- lamen terms remember).
This take us to 50.98. Now it doesn't look that bad at the moment. All we've actually done is made it brighter (or louder). However, now we need to reduce the range. I'm going to do this with the 'Levels'. We need to increase (or technically decrease) it by 11.5% The Levels slider that I'm using has a max of 10 and starts with 1, so I'm going to increase this to 2.15
This has (in a basic way) decreased the dynamic range of the image. In this example, you've lost some of the darker (quieter) detail in the shadows and increased their brightness (volume).
Here's the before and after:
This is more of a subtle difference, and you can make arguments for the changes depending on your preference. However, by making all the shadows brighter, you've lost huge contrast between the bright ray of light and the dark shadows.
Now what happens if we apply the same logic to another picture:
Let's start again by increasing the exposure
Then let's increase the levels
And finally the before and after.
I'm sure most people would agree that the original picture is better. There is an intention to have dark and light areas. It's the same with music, there's an intention to have loud moments and quiet moments. When you compress the dynamic range however, you have less and less space to separate the bright from the dark, the loud from the quiet.
Now feel free to rip me a new one or update/educate me in my comparisons
Pic source:
http://www.superbwallpapers.com/nature/ray-of-sun-through-the-canyon-36249/
http://www.superbwallpapers.com/nature/starry-night-sky-over-the-mountains-14678/
