• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

SINAD Audio Files, Please!

beefkabob

Active Member
Forum Donor
Joined
Apr 18, 2019
Messages
283
Likes
295
#1
The genius @garbulky said we should get recorded files of various SINAD levels and then we can ABX them.

So a song or three. A single tone. A multi-tone.

Then people can listen on their home systems and see if they can tell the difference, ABX.

So feed different models of DAC into an RME or Lynx ADC. Make files in the 60 80 100 and 120 SINAD range.

Maybe a genius can make a website that does it.

Anybody? Anybody?
 

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
7,579
Likes
8,752
#2
Oh you post just the reason I don't like an over reliance on SINAD. I can provide multiple versions of SINAD for you. Some will be ABX'ble and others will not all with the same SINAD. You have to look further than just this one number.

For instance with music, a SINAD of 60 db (with the reason being purely noise) will be obviously audible. Same music with SINAD of 60 db with the reason being only 2nd and 3rd harmonic distortion you most likely could not hear as different. So just knowing the files both have SINAD of 60 db isn't enough to know if it is audible or not.

With music if SINAD is 80 db or more the most common reason for audible differences is rather small frequency response differences.
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
3,764
Likes
6,912
Location
The Neverlands
#3
That would only be useful if the reproducing DAC actually reached at least 120dB SINAD, has excellent multitone response and a frequency response that does not roll-off.

Furthermore the recording side should also have the same properties.

Add to that the amplifiers used should also have 120dB SINAD at loud listening levels.

The result (I predict) is, when FR is flat, the 80 to 120dB SINAD recordings can not be told apart. Some devices with say 30dB to 50dB SINAD may well be reliably detected in proper controled blind tests.
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
3,443
Likes
6,495
Location
Gold Coast, Queensland, Australia
#4
Residual noise is the problem, not THD in my opinion.

When you have gear where the THD is <-110dB, there is none to speak of, and yet some completely non-harmonically related PSU (inaudible) spikes at 60/120Hz can ruin a SINAD number, or in the case of plenty of amplifiers/preamplifiers, a residual noise (hiss) of ~150uV can ruin something with essentially perfect THD.

The result (I predict) is, when FR is flat, the 80 to 120dB SINAD recordings can not be told apart
The recordings themselves maybe, but the "silence" at each start/end of the recording will be the giveaway when amplified. So yes, they can told from one another.
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
3,764
Likes
6,912
Location
The Neverlands
#5
Most DAC's with a SINAD of around 80 have noise floors well below 100dB. The silence before a song could thus consist of mainly mains IF those are unacceptably high.

My personal experience is that when I play at almost uncomfortable loud levels and I attenuate 80dB I hear absolutely nothing.
This explains why I predict 80dB SINAD is enough for home music reproduction.
For professional, sound reinforcement and studio usage 80dB SINAD is a problem.
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
3,443
Likes
6,495
Location
Gold Coast, Queensland, Australia
#6
My personal experience is that when I play at almost uncomfortable loud levels and I attenuate 80dB I hear absolutely nothing.
That's because your poor ears are still ringing and need time to recover.

Go the other way, turn up the quiet bits to hear the differences... :)
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
3,764
Likes
6,912
Location
The Neverlands
#7
That's because your poor ears are still ringing and need time to recover.
Indeed.. so that's why I mentioned you can't tell the 80 -120dB SINAD difference with music. One's actual dynamic range is not big enough in real life situations for this to matter.

Of course when one turns up the volume one can hear the differences between 80dB and 120dB noise floors... but SINAD numbers are rarely determined by noise floor. More often by mains hum or the distortion spikes from the 1kHz tone.

The main reason why I think the SINAD is an indication of technical performance but not the number that says something about SQ when below -80dB IRL (music reproduction) situations. It's just one number in a sea of measurements.
 

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
7,579
Likes
8,752
#8
I'm mostly in agreement with solderdude. Usually an 80 db SINAD product has noise 100 db down and some distortion or maybe hum above that. But all of this points out why SINAD alone isn't a good single number to use. At a minimum you need THD and noise not combined. I've written you would rarely and possibly never with music hear a difference if your gear at the speaker terminals has -70 db of THD and 100 db of SNR. If I'm correct, some 70 db SINAD gear would do the job. And yet other 70 db SINAD gear will not. The problem here is using only SINAD. If you insist on that, then you need at least 100 db SINAD.
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
3,764
Likes
6,912
Location
The Neverlands
#9
The main reason why I reacted to the idea of recording and reproducing at home to listen for differences between the recordings of test tones and music is that when one has a 70dB SINAD DAC and tries to determine how different an 80dB and 120dB SINAD 'recording of another DAC' sounds like it has become a pointless excercise.

To really evaluate a 120dB SINAD DAC recording (reproduced test tones or well recorded (-120dB SINAD ?) music recording one needs an ADC that is better than -120dB SINAD and a DAC that is better than 120dB SINAD as well at home.

My opinion is that 80dB is good enough for music reproduction and can be called 'Hi-Fi' if all (not only SINAD) measurements are good enough to be considered transparent.
 

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
7,579
Likes
8,752
#10
The main reason why I reacted to the idea of recording and reproducing at home to listen for differences between the recordings of test tones and music is that when one has a 70dB SINAD DAC and tries to determine how different an 80dB and 120dB SINAD 'recording of another DAC' sounds like it has become a pointless excercise.

To really evaluate a 120dB SINAD DAC recording (reproduced test tones or well recorded (-120dB SINAD ?) music recording one needs an ADC that is better than -120dB SINAD and a DAC that is better than 120dB SINAD as well at home.

My opinion is that 80dB is good enough for music reproduction and can be called 'Hi-Fi' if all (not only SINAD) measurements are good enough to be considered transparent.
Maybe, maybe not. It would be like speakers only more complex. You can hear with test tones .5% distortion pretty easily even though the speaker is fairly likely to be putting out more than that itself. What you are really hearing is maybe 1.0 % speaker distortion with a clean signal and 1.5% with a dirty signal. Your reference has 1% baked in, and you are hearing the additional distortion on top of that. Now of course if we get the levels of those 20 db apart your gear can cover up the difference. It all would of course come down to harmonic structure etc if you get noise low enough to drop out.
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
3,764
Likes
6,912
Location
The Neverlands
#11
Indeed... but at 80dB SINAD we are already below 0.01% distortion at 1kHz. Ofcourse at 7kHz distortion could well be higher.
At 10kHz the 2nd harmonic is already above the audible range but IM could also degrade the SQ in lower freq. this is why multitone exists.

There is nothing wrong with striving for or desiring best performance though.
 
Last edited:

beefkabob

Active Member
Forum Donor
Joined
Apr 18, 2019
Messages
283
Likes
295
#12
A similar helpful thing would be files with IMD, THD, and other measurements at different levels.
 

beefkabob

Active Member
Forum Donor
Joined
Apr 18, 2019
Messages
283
Likes
295
#13
Also, this is more about curiosity for me. I bought the 2nd tier DAC for my sub because it has XLR output. I bought an all digital streamer because it keeps the signal all digital, which i find elegant. I bought speakers with digital in because they're good speakers. I got XLR all around because I've often run into hums with RCA.
 

pkane

Major Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
1,102
Likes
1,466
Location
North-East
#14
A similar helpful thing would be files with IMD, THD, and other measurements at different levels.
I've been thinking about such a tool -- one that could generate different distortions and apply them to any chosen music track. I think I've seen a website reference that did something similar posted here before. Don't have a link, though...
 

garbulky

Major Contributor
Joined
Feb 14, 2018
Messages
1,137
Likes
432
#15
The genius @garbulky said we should get recorded files of various SINAD levels and then we can ABX them.

So a song or three. A single tone. A multi-tone.

Then people can listen on their home systems and see if they can tell the difference, ABX.

So feed different models of DAC into an RME or Lynx ADC. Make files in the 60 80 100 and 120 SINAD range.

Maybe a genius can make a website that does it.

Anybody? Anybody?
Lol I'm no genius as I'm sure most people here can tell ya! :D


The Procedure
The original digital file. That would be the gold standard - that a person could listen to on their current DAC. It should be made accessible.

Then use a high quality ADC (which has been tested) to record the best measured DACs followed by DACs which measure in lower tiers.
So 120db DAC. 115 db, 110, 105, 100, 96 (CD quality), 90, 85, 80, 75, 70 and 65 (if any can be found).
To take it a step further - if there is a way test it with a combination of an amp and DAC recorded.
Try to level match the recorded levels if possible before recording.

The music could be two categories- one a regular popular piece of music. The other an acknowledged well recorded piece of music that has low distortion in the parts of silence and a good dynamic range preferrably one available in higher resolution. If desired the third piece of music could be a vinyl rip just to see what the threshhold is.However, It mustn't be different pieces of music used with different dacs. Stick to the same choices of music samples over a wide range of dacs for consistency.

Optional: Obviously these categories could be expanded to include genres like heavy metal, electronica and classical if one desires but not necessary since it could be time consuming. A piece of music that appeals to me is to also have some music made with stereo microphones in a low noise non close miced environment - one with no alterations made to it.

It seems relatively straightforward. This could be incorporated into standard testing that Amir does.
The main limitation here I think would be Amir's time. It does take more time to hook these up and run these tests especially with combinations of equipment. But it doesn't have to be Amir. If it's somebody else, the limtiation would be time, equipment, as well as a confirmed high quality ADC. Maybe they could send in their ADC to Amir to test if the measurements are not already published.

What use would this be?
The whole point of this is to allow users to see for themselves what significance the measurement variable of SINAD has when choosing a DAC to them. It's good to quantify exactly how significant listening to less than 110db SINAD is for music use. I am unaware of any studies done on ASR for music. The reason I was given is "what music is the standard?" But it also means I haven't seen any data.

I've seen multiple threads where new buyers come and basically show me that they are using the SINAD chart to entirely base their decisions on DAC buying. They would ignore a DAC with 105 db for a 120 DB DAC.
Some may ignore DACs with useful features like processors like the Emotiva XMC-1 which doesn't get even near 110 db but offer tremendous flexibility like room correction and bass management because of worry of the "sonic impact". So knowing what one is actually getting when looking at SINAD is useful.

How would they do this:
It would allow people with very high performance dacs and/or amps to test for themselves what significant sonic effect (if any) they would encounter if they used a worse performing DAC (for SINAD).

They could perform double blind tests with them and post results or any impressions they care to impart. People with lower performing dacs could look at the results and impressions of these tests and form their opinions.
If a 100 people with a benchmark dac were unable to hear any difference between the original digital file and a DAC recording at 80 db in a DBT, I think that would be some good evidence.

Also people with lower quality older dacs - say one at 95db SINAD may be eyeing a dac upgrade. But if they listen to a 80 db DAC and still hear no difference, then they may wonder why they would want to upgrade their current DAC and focus on what else may be causing dissatisfaction. Perhaps it's their rooms or their speakers which are much more likely to provide larger distortion.
 
Last edited:

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
3,764
Likes
6,912
Location
The Neverlands
#16
Give THIS a try first

or this

or build something like this (from Ethan Winer) and educate yourself.

You will find the bottom limit of what one can hear using music (properly blind tested) isn't anywhere as low as one thinks.
 

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
3,443
Likes
6,495
Location
Gold Coast, Queensland, Australia
#17
but at 80dB SINAD we are already below 0.01% distortion at 1kHz.
Perhaps, perhaps not. You don't know what is noise and what is distortion and which is determining the figure. Most power amplifiers will have residual noise dominating at a SINAD of 0.01%. Most DACs, it will be the THD dominating at -80dB.

As Amir has agreed,he's made his bed with SINAD and there are situations where the numbers do not translate to better performance in real life.
 

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
7,579
Likes
8,752
#18
Lol I'm no genius as I'm sure most people here can tell ya! :D


The Procedure
The original digital file. That would be the gold standard - that a person could listen to on their current DAC. It should be made accessible.

Then use a high quality ADC (which has been tested) to record the best measured DACs followed by DACs which measure in lower tiers.
So 120db DAC. 115 db, 110, 105, 100, 96 (CD quality), 90, 85, 80, 75, 70 and 65 (if any can be found).
To take it a step further - if there is a way test it with a combination of an amp and DAC recorded.
Try to level match the recorded levels if possible before recording.

The music could be two categories- one a regular popular piece of music. The other an acknowledged well recorded piece of music that has low distortion in the parts of silence and a good dynamic range preferrably one available in higher resolution. If desired the third piece of music could be a vinyl rip just to see what the threshhold is.However, It mustn't be different pieces of music used with different dacs. Stick to the same choices of music samples over a wide range of dacs for consistency.

Optional: Obviously these categories could be expanded to include genres like heavy metal, electronica and classical if one desires but not necessary since it could be time consuming. A piece of music that appeals to me is to also have some music made with stereo microphones in a low noise non close miced environment - one with no alterations made to it.

It seems relatively straightforward. This could be incorporated into standard testing that Amir does.
The main limitation here I think would be Amir's time. It does take more time to hook these up and run these tests especially with combinations of equipment. But it doesn't have to be Amir. If it's somebody else, the limtiation would be time, equipment, as well as a confirmed high quality ADC. Maybe they could send in their ADC to Amir to test if the measurements are not already published.

What use would this be?
The whole point of this is to allow users to see for themselves what significance the measurement variable of SINAD has when choosing a DAC to them. It's good to quantify exactly how significant listening to less than 110db SINAD is for music use. I am unaware of any studies done on ASR for music. The reason I was given is "what music is the standard?" But it also means I haven't seen any data.

I've seen multiple threads where new buyers come and basically show me that they are using the SINAD chart to entirely base their decisions on DAC buying. They would ignore a DAC with 105 db for a 120 DB DAC.
Some may ignore DACs with useful features like processors like the Emotiva XMC-1 which doesn't get even near 110 db but offer tremendous flexibility like room correction and bass management because of worry of the "sonic impact". So knowing what one is actually getting when looking at SINAD is useful.

How would they do this:
It would allow people with very high performance dacs and/or amps to test for themselves what significant sonic effect (if any) they would encounter if they used a worse performing DAC (for SINAD).

They could perform double blind tests with them and post results or any impressions they care to impart. People with lower performing dacs could look at the results and impressions of these tests and form their opinions.
If a 100 people with a benchmark dac were unable to hear any difference between the original digital file and a DAC recording at 80 db in a DBT, I think that would be some good evidence.

Also people with lower quality older dacs - say one at 95db SINAD may be eyeing a dac upgrade. But if they listen to a 80 db DAC and still hear no difference, then they may wonder why they would want to upgrade their current DAC and focus on what else may be causing dissatisfaction. Perhaps it's their rooms or their speakers which are much more likely to provide larger distortion.
You could do what you are describing. In a sense when I posted the 8th generation files you were doing that as noise and SINAD decreased with each generation. I could barely detect the difference blind on the 8th generation, but not 6th or 4th. And even then it was a ripple in FR I heard I am pretty sure. If you listen to those and hear the 8th I could make other generations available.

Edit to add, the 1st gen of that had SINAD of 108 db (March Audio DAC+Zen Tour ADC). The final 8th generation had a SINAD of 86.4 db just for reference.
 
Last edited:

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
7,893
Likes
4,866
Location
Riverview FL
#19
I'm thinking the preamp is the weak link in my system (excluding the speakers).

Turning the preamp wide open produces no audible nor in-room measurable noise.

A -90db (1bit) shaped dithered 16bit tone is plainly, though quietly, audible at full volume, without apparent noise, so I'm gonna not worry about it.

Try that.
 

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
7,579
Likes
8,752
#20
I'm thinking the preamp is the weak link in my system (excluding the speakers).

Turning the preamp wide open produces no audible nor in-room measurable noise.

A -90db (1bit) shaped dithered 16bit tone is plainly, though quietly, audible at full volume, without apparent noise, so I'm gonna not worry about it.

Try that.
Your taking all the fun out of it Ray. I've told a number of people who wonder if a quieter this or that would help, "turn your volume all the way up,.........what do you hear? If it is silence or near silence with just a little low level swishing noise don't worry about it". Easiest test we can perform and ears only to boot.
 
Top Bottom