• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

God of SINAD vs. reality we get from most available music files

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
Or with the final fade-out :). Ricki-Lee Jones has something about 30 dB usable dynamic range.
Please excuse my ignorance. Can someone please explain how you determine the usable dynamic range of a recording? The Rickie Lee Jones track has 0dB peaks. So do you just take RMS of the last few 10ths of sec of the fade out (referenced to 0dB or -3dB)? It seems that doing so gives a very high dynamic range so I assume you are doing something else?

edit: so this is what I see on the fade out. The amplitude of the noise looks like it is about 0.00012 or less. Peak of the track is 1. 1/0.00012 = 8333 or about 78dB. I guess this isn't how it is done, though. How do you determine usable dynamic range?
1660068999266.png
 
Last edited:

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,751
Likes
4,633
Location
Sweden - Слава Україні
8 bits (dithered) should sound the same as 16 or 24 bit files in that case. (6 bits needed for 30dB range)
Go ahead and use Google. There are examples of the properly dithered 8-bit music files in the web space together with blind test option against original files. You might get very surprised.
I find this bon mot interesting (which may display my ignorance -- but, hey, at least I am still driven by curiosity!). Mind you, I don't pretend to understand too much about digitization (or whatever one would call this), but resolution in both x and y dimensions is important, isn't it? Is one more important than the other? 'cause there is no doubt (ahem that's probably too strongly worded: let's say I would opine ;) ) that a good old LP record can sound objectively pretty good, but obviously has limited dynamic range compared to other media (and, thanks to the rec/play EQ curve used for LPs, limitations to frequency response, too).

What would (what does) "CD quality" (44.1 kHz) sampling rate, 8 bit audio sound like compared to 14 or 16 bit when converted back to analog?
Correct me if I'm wrong. Bits give an indication of the potential, but what is then measured and determines the quality of the signal/sound is:
FR, noise and distortion.... or?

In other words, number of bits in itself is no guarantee of good signal/sound?
 
Last edited:

mhardy6647

Grand Contributor
Joined
Dec 12, 2019
Messages
11,216
Likes
24,175
No idea. My simple-minded notion is that the number of bits available determines the potential resolution in the y-dimension (voltage).
 

Ra1zel

Addicted to Fun and Learning
Joined
Jul 6, 2021
Messages
531
Likes
1,048
Location
Poland
Correct me if I'm wrong. Bits give an indication of the potential, but what is then measured and determines the quality of the signal/sound is:
FR, noise and distortion.

In other words, number of bits in itself is no guarantee of good signal/sound?
One thing is quantization resolution other is ENOB. I recommend some reading on both.

Edit: while you are at this also explore Johnson–Nyquist noise, to know where are the limits of gear (hint: 24 ENOB is probably impossible at room temperature).
 

Thomas savage

Grand Contributor
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
10,260
Likes
16,298
Location
uk, taunton
I am not working for MQA
OK but ol Bob has been hinding in your basement ( unused cinema) for some time now after all this MQA back lash and abject financial disaster linked to its demise.

Soo.. , you could just ask him init .
 
Last edited:

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,751
Likes
4,633
Location
Sweden - Слава Україні
No idea. My simple-minded notion is that the number of bits available determines the potential resolution in the y-dimension (voltage).
Well, what else could it be? I'm mostly curious if I missed something.:)
One thing is quantization resolution other is ENOB. I recommend some reading on both.

Edit: while you are at this also explore Johnson–Nyquist noise, to know where are the limits of gear (hint: 24 ENOB is probably impossible at room temperature).
Yes. The technology itself::)


But to then measure the quality of the signal/sound. What besides FR, noise and distortion is there?

Okay, absolute phase, time delays, smoothness/deviations between different channels. For speaker directivity, how the sound spreads, that determines how you experience the sound.:)

"In basic implementations, variations in bit depth primarily affect the noise level from quantization error—thus the signal-to-noise ratio (SNR) and dynamic range. However, techniques such as dithering, noise shaping, and oversampling can mitigate these effects without changing the bit depth. Bit depth also affects bit rate and file size."

 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
What would (what does) "CD quality" (44.1 kHz) sampling rate, 8 bit audio sound like compared to 14 or 16 bit when converted back to analog?

That will totally depend on the recording itself and if the 8 bit file is dithered or not (and the type of dither used).
There will be cases where the 8 bit and 16 bit recordings can not be told apart and some recordings where (dither or quantization)noise in silent passages will be the give-away.

I would opine ;) ) that a good old LP record can sound objectively pretty good, but obviously has limited dynamic range compared to other media (and, thanks to the rec/play EQ curve used for LPs, limitations to frequency response, too).

Yes, vinyl can sound pretty good. It has limited dynamic range and there are indeed some practical limitations as well as noise. The frequency response limit (highs) is determined by the needle/cartridge/disc speed/cutting lathe (speed) and cutting head, azimuth. Not by the RIAA curve itself but some RIAA amps may have bandwidth limiting.
In signal fidelity/accuracy vinyl is crap as a medium. Fortunately our brains accept some of vinyl's shortcomings gracefully.
 
Last edited:

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,751
Likes
4,633
Location
Sweden - Слава Україні
I need and can fix the FR by 4dB at 2000 Hz but only if my SINAD gets X% worse.
(hypothetical example)

I think this worse X is quite large for most people. Sensible FR wins, in most cases.:)

Edit:
A theoretical example example, but how do you decide, in practice, what is important to you?:)
 
Last edited:

Ra1zel

Addicted to Fun and Learning
Joined
Jul 6, 2021
Messages
531
Likes
1,048
Location
Poland
For speaker directivity, how the sound spreads, that determines how you experience the sound.:)
The big problem is that how exactly you experience that will vary from person to person. Brain does more processing than any audio device ever will.
 

DanielT

Major Contributor
Joined
Oct 10, 2020
Messages
4,751
Likes
4,633
Location
Sweden - Слава Україні
The big problem is that how exactly you experience that will vary from person to person. Brain does more processing than any audio device ever will.
And that makes this with HiFi fun. Think what a sad world it would be if everyone thought exactly the same about how they wanted it to sound.:)

Note, exactly the same..

Edit:
That being said, ignore speakers. DAC instead. FR, noise and distortion or?
(plus channel balance)
 
Last edited:

Cbdb2

Major Contributor
Joined
Sep 8, 2019
Messages
1,530
Likes
1,485
Location
Vancouver
I think @amirm read your post perfectly. Your first post (above) mentions ELECTRICAL dynamic range, not acoustical dynamic range. That refers to the circuitry, which is why Amir mentioned a "container", and it being "transparent".

Jim
Amir was talking about meters, including software, that measures dynamic range in music, not talking about audio hardware. All I was saying was that its hard to make ones that correlate to our hearing.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,522
Likes
37,050
Please excuse my ignorance. Can someone please explain how you determine the usable dynamic range of a recording? The Rickie Lee Jones track has 0dB peaks. So do you just take RMS of the last few 10ths of sec of the fade out (referenced to 0dB or -3dB)? It seems that doing so gives a very high dynamic range so I assume you are doing something else?

edit: so this is what I see on the fade out. The amplitude of the noise looks like it is about 0.00012 or less. Peak of the track is 1. 1/0.00012 = 8333 or about 78dB. I guess this isn't how it is done, though. How do you determine usable dynamic range?
View attachment 223308
I don't see what is wrong with it other than maybe using an RMS level of it if you can get a clean 1 second to measure with software instead of the graphical way you have done it. Though in principal it is more or less the same idea. You determine the noise floor. The music may or may not fill all of that, but you have the noise floor and max level it has to be working within. I'd like to know how Robert Stuart did it. And I need to dig up a post JJ had about some matlab code he used to investigate it. He didn't find a wide range instead a very restricted range in the few examples he examined.

You can export each sample value in Audacity, but that can be deceptive depending up on how you use it.
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
I don't see what is wrong with it other than maybe using an RMS level of it if you can get a clean 1 second to measure with software instead of the graphical way you have done it.
Thank you for helping me.

I can only get a half second of a consistent level in the section I showed. If I measure that in Audacity I get -93dB RMS. RMS of full scale P-P is -3dB so then we have 90dB dynamic range doing it that way. A significant difference from the 78dB I got the other way. But where did @pma 's 30dB estimate for usable dynamic range come from?
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,522
Likes
37,050
Here is the thread from @j_j where he analyzed some tracks for various bit levels used.

 

Ra1zel

Addicted to Fun and Learning
Joined
Jul 6, 2021
Messages
531
Likes
1,048
Location
Poland
Here is the thread from @j_j where he analyzed some tracks for various bit levels used.

Quite illuminating, shows that most likely by far the most limiting factor in our audio systems are the recordings.
 

Vacceo

Major Contributor
Joined
Mar 9, 2022
Messages
2,634
Likes
2,752
They are already a pain to fix. They want me to ship the monsters to some repair shop in LA as Harman no longer services them. I am sure they just going to swap boards and charge thousands of dollars for each. I don't think they will give me the service manual but I will ask at some point.

Boy do I wish I had some lightweight, module based class D amps.
The crossover between audiophilia and powerlifting is a real and not fun at all. You also tested it with the Genelecs...

This leads me to a technical question: does the rule of "no replacement for displacement" is really as clear as it is commonly stated. You´re an ace of debunking lay intuition, so would it be theoretically possible to build a subwoofer that pushes a relatively high SPL at 20hz in a size that is not a washing machine? If a piece the size of my finger can smoke the power delivery of a class A, will we ever see the equivalent with a subwoofer?
 

danadam

Addicted to Fun and Learning
Joined
Jan 20, 2017
Messages
956
Likes
1,496
edit: so this is what I see on the fade out. The amplitude of the noise looks like it is about 0.00012 or less. Peak of the track is 1. 1/0.00012 = 8333 or about 78dB.
You can switch Audacity to show the waveform in dB scale, after right-clicking on the scale:
audacity-db.png
 

sq225917

Major Contributor
Joined
Jun 23, 2019
Messages
1,360
Likes
1,610
Dac pre and power all around - 120db imd 32 tone, check. Phonostage over -85db thd+noise, check.

That'll do.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,522
Likes
37,050
You can switch Audacity to show the waveform in dB scale, after right-clicking on the scale:
View attachment 223344
I should have mentioned that earlier. Here is a good example from a 2L recording. Both tracks are the same. First 3 seconds. Lower one is linear view and upper is db view. You can see there has been a fade in that would not be at all apparent in the linear mode. You can also select a section as I've done in this view, go to Analyze and choose Contrast Analysis and it will give you the RMS value for the selected segment. In this case it shows -76 db. If you expand the view in db scale you'll find the fade in continues and doesn't end for couple more seconds. At that point the last second is only -65 dbFS which is what I've found for each track of this 2L recording. The peaks for this one are about -2 db so I'd say you have 63 db of dynamic range to work with. If you filtered out everything other than 3-5 khz in this noise, you'll see level around -95 db FS. Is this the area we should decide upon how much range is needed?

1660082385061.png
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
I should have mentioned that earlier. Here is a good example from a 2L recording. Both tracks are the same. First 3 seconds. Lower one is linear view and upper is db view. You can see there has been a fade in that would not be at all apparent in the linear mode. You can also select a section as I've done in this view, go to Analyze and choose Contrast Analysis and it will give you the RMS value for the selected segment. In this case it shows -76 db. If you expand the view in db scale you'll find the fade in continues and doesn't end for couple more seconds. At that point the last second is only -65 dbFS which is what I've found for each track of this 2L recording. The peaks for this one are about -2 db so I'd say you have 63 db of dynamic range to work with. If you filtered out everything other than 3-5 khz in this noise, you'll see level around -95 db FS. Is this the area we should decide upon how much range is needed?
How do you change the minimum when displaying vertical axis in dB? Default for me was -60dB and I can't figure out how to change it. Probably an obvious thing but I can't find the setting.
 
Top Bottom