Understanding Audio Dynamic Range / SNR (Part 1)

[jasonhanjk]

Dynamic range and SNR are actually the same.
If you see both numbers are off by a few dB (eg SNR better result). then the dynamic range is the more accurate one.

 

[Blumlein 88]

Dynamic range and SNR are actually the same.
If you see both numbers are off by a few dB (eg SNR better result). then the dynamic range is the more accurate one.

Actually no.

SNR is a near max signal vs the residual noise with that signal notched out.

Dynamic range is a -60 db signal vs that signal notched out referenced to max level.

Some devices will modulate the noise floor by raising noise level along with increased signal levels. Such devices will have a lower SNR than Dynamic range spec.

So one will not be more accurate than the other, they aren't quite the same thing.

All this is part of the AES-17 specifications. AES-17 is the spec for ADCs and DACs.

 

[amirm]

Dynamic range is a -60 db signal vs that signal notched out referenced to max level.

That is only a measurement standard. When we say "CD has 96 dB dynamic range" we are not talking about any kind of reference to -60 dB, etc.

 

[MC_RME]

Actually no. SNR is a near max signal vs the residual noise with that signal notched out.

Actually no. The signal is not notched out but turned off/muted. In DR it is notched out.

 

[B4ICU]

Dynamic Range: As was well explained by Mr. Julian Hirsch (Stereo Review) in the 70's, it is the range between the loudest sound and the lowest sound. The loudest, might be the loudest we can play, before getting into large distortion figures. Regardless of analog recording, nor SPL.
The lowest, as defined here, is the noise floor. With electronics, it can be relatively low, as Johnson noise, Op Amp's noise etc' with analog sources (vinyl or tape) is the rumble or hiss (-50dB) or CD with S/N of -100dB or better.
Mostly, such dynamic ranges, also need to consider the lowest sound (music) ref. to the noise floor. A 6dB ration, is as good as a tranche line phone of WW-II. So that would be nice to have a decent distance between the two. With digital sources, as CD (or better), the lowest bit (LSB, to none) is
at -96dB. at the lowest effective bits are about 4-6 bits + dither. So S/N below -115 dB is fine.
Another issue, is the fact that we use a serial system: Source, Pre, Power etc'. For such a set, the combined S/N would be the worst out of the list.
If it would be a FR figure it would be a RMS of all FR.
As wrote B4, the environmental noise is way higher than the digital sources or fine SS amplification.

 

[MeMe]

That is only a measurement standard. When we say "CD has 96 dB dynamic range" we are not talking about any kind of reference to -60 dB, etc.

Hi Amir, I'm curious to know if the Professional Amp / HiFi Amp valley of separation is real or not.
In other words, if Pro Amp gear can perform as HiFi or if this is a no-go.
There is some inexpensive used Pro gear waiting to be grabbed if (at least some) if this kind of equipment is suitable for audiophiles.
Thanks!

 

[Schackmannen]

Actually no. The signal is not notched out but turned off/muted. In DR it is notched out.

I thought that for SNR tests the noise is measured with something like a -144 dBFS tone to prevent the DAC from muting its output and giving unrealistically low noise figures, is that not correct?

 

[Newk Yuler]

It's interesting that Paul McGowan is touting a new hour long tour video of his facility that he repeatedly casually points to the Audio Precision analyzers at various work stations.

I find it humorous that Amir's use of an APx555 has made some audio manufacturers decide to buy them so they can show potential customers that they care enough to use them, too. It still doesn't mean they use them to improve the performance of their designs.

"Yes, this fine APx555 is showing me that this piece is delivering our normal level of noise filled audio. QC passed!"

 

[PeteL]

It's interesting that Paul McGowan is touting a new hour long tour video of his facility that he repeatedly casually points to the Audio Precision analyzers at various work stations.

I find it humorous that Amir's use of an APx555 has made some audio manufacturers decide to buy them so they can show potential customers that they care enough to use them, too. It still doesn't mean they use them to improve the performance of their designs.

"Yes, this fine APx555 is showing me that this piece is delivering our normal level of noise filled audio. QC passed!"

I think it's fair game here to make fun of some manufacturers marketing approach or engineering prowess, I have no gripe with this. But Quite honestly you are giving Amir's way too much credit here.

I don't see how you can bring a product to completion without measuring equipment, it's a tool, that's all and they were present in every manufacturers labs long before the existence of this website. Now they may not use it the same way, they may not put values on the same metric, but McGowan team or others no matter how we may find it humorous, are still electronic designers (and yes have knowledge that most here don't have)
What we may give credit for this site could be a will to publish their measurment, or to focus on objective performance, that I can give, but assuming they would buy this for show or that they wouldn't use it, sorry, anybody has to be able to see what's going on, and to analyse stuff, even McGowan.

 

[MC_RME]

I thought that for SNR tests the noise is measured with something like a -144 dBFS tone to prevent the DAC from muting its output and giving unrealistically low noise figures, is that not correct?

https://www.ap.com/technical-library/signal-to-noise-ratio-snr-dynamic-range-and-noise/

 

[B4ICU]

Mr. Lambda
3. I assume that you might look at the wrong scale: https://www.healthlinkbc.ca/health-topics/tf4173
Please pay attention to the remark: (sounds above 85 dB are harmful). I'm a wear that today amplified music reach painful damaging levels.
After been exposed to such levels: 150-160dB or 135dB for some period of time, your hearing is damaged permanently. After, I doubt any ability to
"Hear high pitch signal way under the mostly pink/brown background noise". I try to avoid such hearing punishment.
Nevertheless, none of those extreme levels can be really reproduced with a home sound system. Even with music like Mahler 1st symphony's last track. Others, like a jet fighter takes off, or a jet fighter engine is test running at full power for some time, is impossible to recreate or endure for long.
So after all, the human hearing (try out Fletcher Munson Curve) is limited between the lowest sound observation level (1dB) and the damaging levels of 85 dB. A single try with severe damage does not counts.
A CD dynamic range is: 96 dB (Log x 20 2^16 = 96 dB).
Analog sources without dbx or other NR systems are at 50-60 dB.
Just like a THD under 0.1% is to prove that the design is excellent. No one can tell the difference between 0.1% to 0.0001%.

 

[MrPeabody]

S/N ratio is concerned with the quality of a signal, not the medium or the encoding format used to carry the signal, whereas dynamic range is concerned with the limitations of the medium or the encoding format.

With respect to the medium or the encoding format, the concept of dynamic range is predicated on the existence of both maximum and minimum signal levels intrinsic to the medium or the encoding format. This is more straightforward in the case of digital encoding (PCM specifically) as compared to analog media, because with PCM, the weakest signal that can be encoded is the value for the least-significant bit, and the strongest signal that can be encoded is the value corresponding to all bits set. This is easily calculated, and a very accurate approximation is obtained by simply multiplying 6.02 dB by the quantity of bits. [The approximation is very accurate so long as the quantity of bits is great enough for PCM to be useful for audio, but it is nevertheless an approximation. The true, exact value is the ratio (2^N -1)/(2^1 - 1) = 2^N -1 = 65535 for N = 16. Of course you have to take the base-10 log of this number and multiply 20 (20 because the encoded value corresponds to voltage). The reason for subtracting 1 is that one of the values that can be encoded in N bits is 0, thus the largest value is 2^N -1. But the difference is altogether insignificant so long as the quantity of bits is great enough to be useful for audio.]

With analog media and modulation schemes, the concept of dynamic range isn't meaningful unless there is a hard limit on the maximum value that can be represented and also a non-arbitrary way to identify a standard value for the noise floor. When people talk about the dynamic range of vinyl, this is based not so much on any theoretical limitation of the medium so much as empirical assessment of the S/N for whatever is recorded on the vinyl. The best S/N ratio that is possible, as determined empirically, is deemed to be the dynamic range of the medium. This seems entirely reasonable to me, however it also seems worth noting that the way dynamic range for an analog medium such as vinyl is obtained is fundamentally different from the theoretical method that applies to a digital encoding format where the maximum value and the least non-zero value are absolute.

 

[Newk Yuler]

I think it's fair game here to make fun of some manufacturers marketing approach or engineering prowess, I have no gripe with this. But Quite honestly you are giving Amir's way too much credit here.

I don't see how you can bring a product to completion without measuring equipment, it's a tool, that's all and they were present in every manufacturers labs long before the existence of this website. Now they may not use it the same way, they may not put values on the same metric, but McGowan team or others no matter how we may find it humorous, are still electronic designers (and yes have knowledge that most here don't have)
What we may give credit for this site could be a will to publish their measurment, or to focus on objective performance, that I can give, but assuming they would buy this for show or that they wouldn't use it, sorry, anybody has to be able to see what's going on, and to analyse stuff, even McGowan.

The thing is I see what's going on when Paul goes out of his way to point out every AP he passes in the tour video. To the general audience it's a statement of using fine analyzers but to a sceptical ASR regular it can look like a subtle comment about something directed here. Paul is a showman and salesman who takes advantage of his super friendly, wise, grandfatherly image to sometimes make make subtle comments that aren't really directed at his general audience. He doesn't start pissing matches because he knows there is nothing good to be gained from it. He instead makes very subtle comments and references in his videos and daily blogs that counter things that are critical of him and his equipment. So when he points to every one of his APs it looks like a purposeful reference to his awareness of the ASR criticisms. I dunno, I suppose that's a stretch but I see it that way because I've subscribed to his email blog for a few years and when he comes into focus here there is almost always something subtly mentioned and usually repeated several times to counter it. He just never begins to get aggressive or obvious about it, which is smart. And there is always something seems slightly devious about it when you know how to look at it which supports doubting how genuine he is about other things he says. He's spent his career building the super friendly charismatic showman-salesman persona that sells his elaborate hardware. He's certainly a wise and intelligent man worthy of the respect of his clever showmanship.

I apologize if this looks like a feeble attempt to derail the OP. I have nothing but the deepest respect for Amir and his endeavors. This just came to mind because we're talking about the AP and its functions again.

 

[anchan]

I second this question. All noise is definitely not created equal. To say we want +10db over a bitrate's dynamic range is a great rule for modern equipment, but doesn't really address a few things.
With some equipment, you might have some mains related noise. Turns out that a -80 dBV peak at 60Hz is a heck of a lot better than that same level at 2 kHz

Amir, could you address the advantages/disadvantages of applying A-weighting? Or perhaps in your reviews you could overlay the fletcher munson curve to show relative impact of noise to our (human) ears? Would love to hear some thoughts.

Why not apply a weighting like (A) to the noise measurement?
Because two peaces of equipment with the same 120dB SNR maybe have different SNR in dBA.

 

[stunta]

It might be a good time to revive the room measurements for dummies threads and make them into a YouTube series.

Yes!

These videos are gold mines, @amirm. Thank you and please keep them coming. These clarify how to interpret your measurements and make more informed purchasing decisions.

I wish I had these resources when I started getting into high-end audio - would have saved me thousands of dollars and countless hours.

 

[amirm]

Thanks for the feedback you all. My current thought is to pick each measurement and demonstrate it in real-time. Then I can create a playlist for someone to easily find them all. Original idea was to do them all in one video but I realized a focused chunk at a time like this one works better. Keeps the videos shorter and allows a deeper dive into what each one really means.

The text version of these would become a book! So just not in the cards as far as making the time to write it.

 

[TheBatsEar]

What i don't get is where the noise comes from in simple devices. For example, if you have a mobile DAP, you don't have a complex power supply, in fact, the amount of parts is reduced as much as possible. Less parts, less noise, right? Shouldn't that mean you have less noise, or the chance for higher dynamic range?

 

[amirm]

Even a simple resistor generates noise. You don't need a power supply to create noise. But note that that the DAP also has a power supply to convert the battery voltage to the values it needs.

 

[luft262]

Awesome video Amir! Thank you for educating us all. I'm sorry if I'm a slow learner, but am I right in saying that as you turn the volume down the dynamic range is getting lower so that if you're usually listening to music softly there is less need for equipment/electronics with a high (120dB/20bit) dynamic range? Thank you for your expertise!

 

[luft262]

Does anyone understand what the dynamic range database is trying to portray?

https://dr.loudness-war.info/

For example, if an album is listed with a maximum dynamic range of 15 does that mean it has 15dB of dynamic range or 15 bits or what???

Thanks!