• 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

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,525
Location
Seattle Area
i would be happier if our beloved SINAD was the maximum SINAD measured on a range of voltage (0.5 to 2V) and frequency (5hz to 20khz).
I show the latter in every review. The former is also there for any DAC with variable output.
 

Cars-N-Cans

Addicted to Fun and Learning
Joined
May 19, 2022
Messages
819
Likes
1,009
Location
Dirty Jerzey
For headhone amplifiers (high sensitivity IEMs/ear/headphones) and amps powering high efficiency horn speakers there is a real need to have really low noise numbers.
These are also included in SINAD. Fortunately Amir also shows S/N (at 2V or 4V) and SINAD at 50mV or 5W giving us that info.
I would definitely agree that noise is one of the most important aspects since there is nothing playing to mask it, and it doesn't correlate to the music. The more efficient a speaker or headphone is, the louder the noise will be, and its easy to get burned (and I have been) when switching to low impedance speakers that have fairly efficient tweeters, and like you there is nothing worse than mains hum and harmonics, but hiss is up there as well. As far as distortion goes, I have spent more time than I ought to have squinting into speakers and headphones trying to see how much distortion can be heard, and it seems, for me at least, once its below about -50 dB, its out of the picture perceptually.
 

Rja4000

Major Contributor
Forum Donor
Joined
May 31, 2019
Messages
2,682
Likes
4,220
Location
Liège, Belgium
It encourages mediocracy in equipment design which results in no benefit to the consumer.
I guess SINAD ranking may be considered as an activist's weapon.

More detailed data are available in the reviews to help one's choice, if that's what one's seeking for.
 

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
5,844
Likes
5,787
I show the latter in every review. The former is also there for any DAC with variable output.
And really usefull determining the actual SINAD of the device one has at his normal listening volume,specially in a simple rig,dac hooked straight to power amp for example.
 

Lambda

Major Contributor
Joined
Mar 22, 2020
Messages
1,785
Likes
1,519
Fighting SINAD is like fighting with a restaurant to not have clean dishes. It is not like your food would be cheaper if they served your food on dirty ones.
Sure!
But we have to remember that super clean dishes don't mean super good food.

I show the latter in every review. The former is also there for any DAC with variable output.
That’s super nice!
May i make the suggestion staying with your dirty dishes analogy.
We generally don’t care about the cleanest dishes we get at a restaurant but the dirties?

In addition to having rating bar graph measured at standard at 1khz 2/4V measurement it would be nice to have a "Worst cases" "sinad"*
as well as best cases SIAND.

1659869600531.png


*I don't know how Worst cases would be defined. Maybe multi tone or at low volume?

For example the topping-d70s-mqa. its exelent with an SINAD of 120dB at the standard measurement
But jitter is almost at -100dB and and THD+N at 500Hz is below -100dB
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
For me, the interesting part of this thread is not the debate over how much SINAD we need with our electronics, it is understanding the use of dither with 16 bits. It's worth messing around with it yourself. For example, make an un-dithered 1KHz file like the 2nd plot in post #1, but with a notched out fundamental. If you turn up the volume you can hear just how nasty the quantization distortion sounds (even though this is normally masked and low level).

It is easy to make tracks of silence with triangle and shaped dither with Audacity, and turn those up to hear how they sound (easy to add a lot of gain on the fly in DeltaWave). Or see if you can hear anything from your system as you use it when you play the dithered silence (in my case hear it very faintly with my ear to the speaker when playing at full volume).

Anyway, dither is pretty cool stuff and can be handy to have some understanding of it. For example, I mostly listen with Qobuz through a Sonos Connect. The Connect just truncates any 24-bit files (and also makes a mess of 48KHz files), so in that case I have the quantization distortion issue. However, I can play any Qobuz files with Roon, upsample for DSP and then downsample to 16/44 with dither. Doing this tames the Connect's faults and I couldn't be happier with the final 16 bit sound and feel no need to upgrade to a 24 bit streamer.
 
Last edited:

Ron Party

Senior Member
CPH (Chief Prog Head)
Joined
Feb 24, 2016
Messages
411
Likes
559
Location
Oakland
I believe that is this one, which I have, and I think his estimate was 72 db, but I may be remembering it wrong. My estimate is around 70-72 db. My estimation was to find silence without music playing and during which no gain adjustment was made and see what level the noise is.

Here's Arny's post on the matter:

The widest dynamic range commerical recording I know of is some fairly recent Minneapolis Symphony stuff on the BIS label, delivered as a dual-layer SACD. I have a SACD player and in my tests, the CD layer is mastered the same as the SACD (so its a fair comparison) and both show an actual musical dynamic range of about 85 dB.

I believe this is the disc to which he refers.

He also wrote that prior to the BIS recording, "the high water mark for dynamic range" that he found was Ricky Lee Jones's first album, which came in at "about 73 dB dynamic range."
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,525
Location
Seattle Area
I don't know where Arny got his information. Analyzing true dynamic range requires distinguishing pure noise from music in the noise. This requires statistical analysis of the noise spectrum and correlating it with the music. I have never seen a commercial tool for this or any post from Arny indicating he knew how to do this. I am pretty sure I had this very argument with him.

Maybe he was just using an SPL meter and looking at the peak vs quiet segments. That is not a good method as the peak levels are level/amplification dependent.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,525
Location
Seattle Area

Thomas_A

Major Contributor
Forum Donor
Joined
Jun 20, 2019
Messages
3,423
Likes
2,407
Location
Sweden
If you max out peak music 115 dB in your room minus room noise 30 dB you have 85 dB left. Hearing below the noise floor (10 dB) gets you down to 95 dB. So equipment would have higher SNR than that.

In ideal maxed out situations?
 

Ron Party

Senior Member
CPH (Chief Prog Head)
Joined
Feb 24, 2016
Messages
411
Likes
559
Location
Oakland
Amir, I actually have those SACDs on my server. Is there an easy way I can test them for dynamic range?
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
How I've measured noise in recordings is very simple. I find a few seconds just before the music starts where you can measure the noise level in software. Many recordings they fade in or out of each track and you cannot do that. Quite a few classical recordings will have places that isn't done and you can measure it. The music itself may be more or less dynamic than that, but the noise in such times is the lower floor noise limit in those conditions. The best I've come across are near -70 db and not many get that low. 55-60 db is more common.

How should we consider noise or dynamic range? I've commented when people use an SPL meter to measure their listening room you should take into account that our hearing is most sensitive to 3-5 khz. A listening room with a 35 db SPL noise level will be 10 db SPL or less in that 3-5 khz range. You can create low level tones or band limited noise and hear it in those frequencies at those low levels. If you have good hearing you hear into the noise enough to hear right near 0 db SPL.

We don't measure gear that way however. A-wtd measures are a poor attempt to somewhat compensate for our hearing sensitivity that varies with frequency. Stuart and Craven when they say we need 18 bit for music are applying those curves similar to Fletcher-Munson curves. Like this one from the paper I linked to earlier in the thread. I also have some questions about their statistical surveys of thousands of recordings. I don't know enough about how they did that to know if I agree with them or not.

1659911456740.png

So should we do the same with gear measurements? Some gear with pretty poor Sinad could look a lot better if evaluated this way. I understand why Sinad is used. If gear is good enough on Sinad it means you can confidently say it is transparent under any conditions. And Amir's reviews give us plenty of ways to parse apart which are maybe ok or not when Sinad isn't great. And plenty of gear is capable of Sinad results that are great without costing a bundle you can ask why would you accept anything less? Someone not knowing these nuances however end up just looking at Sinad and sometimes say "I've got a gear that isn't great on measures, but it sounds fine to me." Or sometimes you have older gear which isn't great on Sinad yet if you spend money, even a little, on one of the great new DACs you might find it sounds no better than before. If I have a 10 year old $2000 DAC that is okay, spending $200 for no improvement is not worth doing.

So Sinad can give a clear dividing line where we know the gear is not a problem. And a line where we know it is a problem. There is a very large gray area where you'll need to delve further into issues to say if something is fine or not fine. As for recordings, while not into LP's anymore, the good ones on good gear are close enough to let you know we don't have to have SINAD greater than 70 db to sound good. Maybe there are some recordings better than 70 db, but the overwhelming majority don't get that far in an overall sense.

Should we go for max everything taking into account info from Stuart and Craven, and the findings of Fielder at Dolby. No less than 125 db SPL capable gear with no less than 115 db dynamic range. I'd love to have it, but not having the funds of the Skywalker ranch it is going to be hard to come by.

So does anybody have any better info on the analysis Stuart and Craven did on commercial recordings? I remember JJ posted some matlab code that would do some analysis like that at one time spitting out bit level histograms and other info. JJ's findings made him lament how little dynamic range most recordings have.
 
Last edited:

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,525
Location
Seattle Area
Amir, I actually have those SACDs on my server. Is there an easy way I can test them for dynamic range?
Not that I know of. I only know some companies like Meridian having the tool. It is probably not hard for someone to write it in Matlab or something. I just haven't spent the time doing it.
 

PierreV

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
1,437
Likes
4,686
Assuming uncorrelated noise and distortion, then you can root-sum-square the noise contribution of devices in a chain. For example, start with a device having -100 dB SNR. Now cascade another device with -110 dB SNR, with everything at unity gain. The resulting SNR will be about -99.586 dB, meaning 10 dB lower SNR has cost about 0.5 dB in SNR.

Now imagine you upgrade your -110 dB DAC -100 dB amp combo with the ultimate -123 dB ASR champion...
You _almost_ gain 0.40 dB. Probably not what people jumping on the latest and greatest expect.
On the other hand, if you stick with the -110dB DAC and upgrade to a -105dB amp, you end up with more than 4 dB gain. Ten times better.
And let's not even consider speakers...

Upgrading the weakest link is what makes most sense, always, and even more so if one aims for high SNR.

I get the feeling that while subjective audiophiles often worry about what doesn't exist, objective audiophiles often worry about what doesn't matter.
 

jhaider

Major Contributor
Forum Donor
Joined
Jun 5, 2016
Messages
2,823
Likes
4,517
As far as I can tell, none of the people complaining about SINAD have pure motives (not addressing you but OP). They start with being annoyed with the notoriety that it has gotten and then go from there. As I explained, it is a silly armor to wear. It encourages mediocracy in equipment design which results in no benefit to the consumer.

Is that actually the case though? One objection to the focus on SINAD is the opportunity cost. Many newer companies seem to be chasing SINAD but not providing the user the ability to improve actual or perceived fidelity - e.g. bass management, parametric EQ, loudness compensation, intuitive tone controls, crossfeed for headphones, etc. Some rebalancing of priorities would lead to more interesting and useful gear.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,376
Likes
234,525
Location
Seattle Area
Is that actually the case though?
It absolutely so. THD+N and hence SINAD has been around for decades. How come these protests had not come about until we as a site became so popular?

One objection to the focus on SINAD is the opportunity cost. Many newer companies seem to be chasing SINAD but not providing the user the ability to improve actual or perceived fidelity - e.g. bass management, parametric EQ, loudness compensation, intuitive tone controls, crossfeed for headphones, etc. Some rebalancing of priorities would lead to more interesting and useful gear.
Those are not their skillset. You are talking about signal processing and embedded software development. It is way outside of the comfort zone of these and frankly, many audio companies. Adding such features also makes the product more unique and more expensive and hence potentially smaller volumes. RME ADI-2 DAC is one such product with those characteristics.

So they do what they are good at which is optimizing electronic performance. Where they need or should go beyond that, is something they need to figure out.
 

maverickronin

Major Contributor
Forum Donor
Joined
Jul 19, 2018
Messages
2,527
Likes
3,308
Location
Midwest, USA
One objection to the focus on SINAD is the opportunity cost. Many newer companies seem to be chasing SINAD but not providing the user the ability to improve actual or perceived fidelity - e.g. bass management, parametric EQ, loudness compensation, intuitive tone controls, crossfeed for headphones, etc. Some rebalancing of priorities would lead to more interesting and useful gear.

The lack of products integrating useful DSP is leaving a lot on the table, but I still think we need both. Plain old high SINAD is pretty useful on it's own for avoiding hiss when using pure digital volume control and for DSP headroom.

Also firmware/software usually ends up being the hardest thing to do right. I don't see companies like Topping or SMSL successfully pivoting to provide those kinds of products even if demand did shift so I don't think there is any opportunity cost.

I agree it's lamentable that products like the RME ADI-2 DAC and Qudelix 5K are practically miracles, essentially unique in the market, but I don't see how Amir choosing a different format for his reviews could have changed that.
 

jhaider

Major Contributor
Forum Donor
Joined
Jun 5, 2016
Messages
2,823
Likes
4,517
It absolutely so. THD+N and hence SINAD has been around for decades. How come these protests had not come about until we as a site became so popular?

I may be speaking only for myself here, but I expect many if not most ASR participants and readers could not have told you what "SINAD" was before you adopted it as your shorthand figure of merit here. And I don't know about "protests" because I don't really follow that stuff. Electronics are in the main boring to me unless they actually solve a real problem. The largest fidelity problems to solve in home and personal audio electronics today are, respectively, bass management + modal equalization, and frequency response equalization. These simply tower over any other potential fidelity issue. So, for any integrated product* "how does it address one or both of those?" is just a rephrase of "why should anybody care about this thing?"

*I'm using "integrated product" as shorthand for an audio component designed to fulfill multiple functions, or be the only processing link in the signal chain.
Integrated product examples: integrated amp/AVR/AVP, DAC/headphone amp, wireless speaker, or DAC with a big volume knob on the front
Not integrated products: analog amplifier (e.g. Buckeye or Benchmark), source component (e.g. AppleTV or digital/analog disk-spinner), or 2-channel DAC that makes it hard for someone to touch it screw up your gain structure (e.g. Topping E50).
One can argue either way about a home audio (digital or analog) preamp or wired powered monitor - it's easy enough to add the necessary capability elsewhere in the chain with such products.

Those are not their skillset. You are talking about signal processing and embedded software development. It is way outside of the comfort zone of these and frankly, many audio companies.

Excuses just bore me to tears. Developing "signal processing and embedded software" is "in the major" of every single integrated audio electronics designer or manufacturer today. That's where the audible improvements lie. So there's no excuse for them not to develop, or seek out, the talent required to make things that actually solve problems instead of just pumping out the next darling me too pointless who cares whatever with modest improvements in inaudible specifications compared to last product cycle's darling me too pointless who cares whatever. That, simply put, is the job!

Adding such features also makes the product more unique and more expensive and hence potentially smaller volumes. RME ADI-2 DAC is one such product with those characteristics.

I think your "more unique" is my "worth a damn." Also, I think the "more expensive" argument fell out the window like a Russian dissident when Quedlix 5K came onto the market. To my knowledge, neither RME or Quedlix are particularly large companies. Maybe Quedlix has a parent I don't know about? WiiM (maybe a subsidiary of a large company, I don't know) also claims they will add EQ capabilities to their little $100 hockey puck sized streamers.

For that matter, look at miniDSP. They're certainly not a large company. Nor are their products particularly expensive in the scheme of things. They just decided to focus on attacking what matters.

So they do what they are good at which is optimizing electronic performance. Where they need or should do beyond that, is something they need to figure out.

I agree with that. However, I submit that one role of someone who has amassed some influence through his own diligence, commitment, and smarts is to use that influence to guide these companies to skate to where the proverbial puck is going to be rather than where it was.
 
Last edited:
Top Bottom