Audibility of SINAD differences?

[Sokel]

After a gazillion of tests with all kinds of distortion (including IMD,TIM,etc) I now care only about the noise part of SINAD.
Seems like the old 0.1% THD limit in the pro world is legit and plenty.
Something must be seriously broken or deliberately made to exceed that.

I find gain structure and noise far more important.

 

[DanielT]

Generally about distortion, audibility and a bit about the research around that, I found this on another forum. The reasoning is based on speakers, which is reasonable because they are usually the biggest culprit in the drama.

As for the basic question, there are mostly older reports available in the field if you are interested in more elaborate science. In my opinion, these works suffer badly partly from the rather inadequate recording and playback systems of the time and partly from not having divided the listeners into insensitive and sensitive groups and have therefore arrived at relatively high limits of audibility.

Since the majority of listeners are quite insensitive to distortion, this is not a priority area and no one will therefore invest the resources required to dig deep. In addition, the insensitivity unfortunately increases via listening to today's torn and thus very distortion-dense productions. It's not just the lack of dynamics that makes the audiophile's ears bleed.

In general, however, it can be said that it seems that sensitive listeners under optimal conditions can approach the hearing threshold, but that you (of course) cannot fall below it (I would not attach too much importance to the subliminal speculations). In practice, however, the background level in the room in the frequency band where the distortion components appear will constitute the final limit.

Masking means that harmonic components often become completely inaudible, while intermodulation components often appear very clearly. Even perhaps less well-known issues like Doppler Distortion (FM Distortion or a form of IMD) have a greater impact on the sonic characteristics than one might think - especially with some modern drivers in a two-way configuration, where linearity is good but the radiating area is small, which leads to high diaphragm velocities already at moderate sound pressure levels. Even turbulence modulation noise from bass reflex ports can be clearly detectable as a particular kind of rawness or indistinctness in the sound.

Anyone who has embraced the use of truly low-distortion speakers knows what this means - going from around 0.3-0.5% THD in the midrange and treble to around 0.05-0.1% THD clearly reduces a grittiness that you don't always be aware of before the switch. A warning should be issued here, as there is no going back after hearing what it should sound like.

It is also interesting that deliberately rather "dirty" popular music recordings also sound cleaner, or perhaps rather more distinct, even though the raw sound even increases, i.e. that the listener hears exactly the distortion that was intended. The difference can sometimes be strikingly large even here, which probably surprises many.

faktiskt.io • Visa tråd - Intermodulationsdistorsion på gott o ont.

 

[Matt_Holland]

Generally about distortion, audibility and a bit about the research around that, I found this on another forum. The reasoning is based on speakers, which is reasonable because they are usually the biggest culprit in the drama.

As for the basic question, there are mostly older reports available in the field if you are interested in more elaborate science. In my opinion, these works suffer badly partly from the rather inadequate recording and playback systems of the time and partly from not having divided the listeners into insensitive and sensitive groups and have therefore arrived at relatively high limits of audibility.

Since the majority of listeners are quite insensitive to distortion, this is not a priority area and no one will therefore invest the resources required to dig deep. In addition, the insensitivity unfortunately increases via listening to today's torn and thus very distortion-dense productions. It's not just the lack of dynamics that makes the audiophile's ears bleed.

In general, however, it can be said that it seems that sensitive listeners under optimal conditions can approach the hearing threshold, but that you (of course) cannot fall below it (I would not attach too much importance to the subliminal speculations). In practice, however, the background level in the room in the frequency band where the distortion components appear will constitute the final limit.

Masking means that harmonic components often become completely inaudible, while intermodulation components often appear very clearly. Even perhaps less well-known issues like Doppler Distortion (FM Distortion or a form of IMD) have a greater impact on the sonic characteristics than one might think - especially with some modern drivers in a two-way configuration, where linearity is good but the radiating area is small, which leads to high diaphragm velocities already at moderate sound pressure levels. Even turbulence modulation noise from bass reflex ports can be clearly detectable as a particular kind of rawness or indistinctness in the sound.

Anyone who has embraced the use of truly low-distortion speakers knows what this means - going from around 0.3-0.5% THD in the midrange and treble to around 0.05-0.1% THD clearly reduces a grittiness that you don't always be aware of before the switch. A warning should be issued here, as there is no going back after hearing what it should sound like.

It is also interesting that deliberately rather "dirty" popular music recordings also sound cleaner, or perhaps rather more distinct, even though the raw sound even increases, i.e. that the listener hears exactly the distortion that was intended. The difference can sometimes be strikingly large even here, which probably surprises many.

faktiskt.io • Visa tråd - Intermodulationsdistorsion på gott o ont.

Thanks for this. Very useful.

Yesterday I was at my brother’s apartment where he has quite a nice home studio setup. His mix/mastering monitors are Barefoot Footprint 02 and they were a fantastic reminder of the benefit of a very low distortion speaker. The last similar speaker I listened to was Neumann 310 and the Barefoots reminded me of their sheer clean, fatigue free and hugely resolving sound.

I need to upgrade my Revel M16 soon!

 

[DanielT]

Thanks for this. Very useful.

Yesterday I was at my brother’s apartment where he has quite a nice home studio setup. His mix/mastering monitors are Barefoot Footprint 02 and they were a fantastic reminder of the benefit of a very low distortion speaker. The last similar speaker I listened to was Neumann 310 and the Barefoots reminded me of their sheer clean, fatigue free and hugely resolving sound.

I need to upgrade my Revel M16 soon!

The word of warning in the quote , A warning should be issued here, as there is no going back after hearing what it should sound like, is probably mostly aimed at the listener's wallet.

I've listened to low-distortion speakers that also had high SPL, so they most likely would have passed a tough compression test with flying colors. Then I listened to my speakers at the time and the most obvious thing was that I didn't want to play at very high volume for a long time. A volume level with my speakers that was LOWER than the low distortion speakers I listened to and could enjoy longer listening sessions with.
Unfortunately not for too long , the low distortion speakers,because I heard them at a DIY fair.

Nowadays, I live in an apartment where it is quite audible between the apartments, so for reasons of not disturbing the neighbors, I rarely play at such a high volume. But there are headphones and then I'm back to square one again with requests for low distortion, at reasonably high SPL.

Otherwise a general tip. If you don't need SPL monsters, electrostatic speakers can have really low distortion. For example:

The Quad was always a low-distortion speaker—I remember being with Martin Colloms when he measured an ESL-63 and found that it produced 0.1% THD or less over most of the audioband! Fig.10 shows the spectrum of the ESL-2805's output while it reproduced a 1kHz sinewave at 86dB SPL at 1m. The only harmonics that can be seen are the second, at –72dB (0.025%); the third, at –76dB (0.015%); and the eighth, at –78dB. And while a subharmonic can be seen at 500Hz—as I said in my Richard C. Heyser Memorial Lecture to the Audio Engineering Society in October 2011 (footnote 1), panel speakers behave in a mathematically chaotic manner, which leads to the production of subharmonics—this lies at –70dB (0.03%). The Quad ESL-2805 is considerably more linear than many tube amplifiers!

Quad Reference ESL-2805 loudspeaker Measurements

Sidebar 3: Measurements

www.stereophile.com

 

[antcollinet]

I've always wondered: by now there's a decent amount of rather complete measurements (so SINAD and IMD etc) on ASR and other sites. Did anyone perhaps already take this dataset and try to quantify these relationships? I.e. the correlation between SINAD and the other measurements for instance. Would allow to get an estimate of SINAD vs chance to perform badly elsewhere. Doesn't mean there can't be outliers, but then again the chance for those could be quantified.

My wild guess would be that starting at a certain threshold for SINAD, all the other measurements would also be fine. Not necessarily because some kind of fixed physical relationship between them, but raher because manufacturers striving for a very good SINAD also take care of multi-tone and IMD etc.

Sinad has pretty good correlation with other measurements. That means a good Sinad score is a good indication that other measurements will also score well - but it is not guaranteed. As you say, there are outliers.

Sinad is a good metric to shortlist candidates - but then it is necessary to look at the other measurements to be sure.

 

[ferrellms]

I guess this is a question for Amir.
I’m curious if there’s a correlation between a specific amount of SINAD performance and an audible difference between the amps?

Don’t know if I phrased that right but would an amp within a SINAD of 70 sound audibly different than one of 100 or would an 80 sound different?

Try this document on SINAD audibility

SINAD Audio Audibility Analysis

SINAD in Audio Electronics: Technical Significance and Audibility in a Living Room Context 1. Introduction Overview: The specification SINAD, an acronym for Signal-to-Noise and Distortion ratio, frequently appears in the technical documentation and reviews of audio electronic components such as D...

docs.google.com

 

[Hayabusa]

Try this document on SINAD audibility

SINAD Audio Audibility Analysis

SINAD in Audio Electronics: Technical Significance and Audibility in a Living Room Context 1. Introduction Overview: The specification SINAD, an acronym for Signal-to-Noise and Distortion ratio, frequently appears in the technical documentation and reviews of audio electronic components such as D...

docs.google.com

Nice doc!

 

[PristineSound]

Below is from ChatGPT



The thresholds for THD+N audibility come from a mix of peer-reviewed research, audio engineering standards, and decades of subjective and objective testing in the audio community. Here are key sources and reference points:

---

1. AES (Audio Engineering Society) Papers​

• Benjamin & Gannon, AES Paper 4855 (1998):
  Found that THD below 0.1% is generally inaudible with complex music signals. They used listening tests with trained listeners and various levels of harmonic distortion.
  • Source: AES E-Library

---

2. Floyd Toole / Sean Olive (Harman Research)​

• In studies and books like "Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms", Toole emphasizes that well-designed electronics with THD below 0.1% are unlikely to be audibly distinguishable in blind tests.
  • Book: ISBN 9781138921368

---

3. Douglas Self​

• In "Small Signal Audio Design", Self suggests that THD+N below 0.01% is transparent for audio applications, and that most distortion in that range is dominated by noise.
  • Book: ISBN 9781138850071

---

4. ITU and EBU Recommendations​

• ITU-R BS.1116-3: Specifies conditions for subjective listening tests and implies that extremely low distortion (e.g., <0.1%) is required for transparency under controlled conditions.

---

5. Historical and Industry Practices​

• High-end DACs and amplifiers routinely aim for THD+N below 0.001% as a standard of "inaudibility", even if that level is often overkill. This is often more about measurement excellence than audible benefit.

---

Summary Table:​

THD+N Level	Audibility Summary	Sources
>1%	Clearly audible in most content	AES, general consensus
~0.1%	Possibly audible in test tones, not in music	AES Paper 4855, Toole, Self
<0.03%	Generally inaudible in music	Douglas Self, Harman studies
<0.01%	Transparent and overkill for most listeners	Industry standard, Hi-Fi tests

 

[Chrispy]

Below is from ChatGPT



The thresholds for THD+N audibility come from a mix of peer-reviewed research, audio engineering standards, and decades of subjective and objective testing in the audio community. Here are key sources and reference points:

---

1. AES (Audio Engineering Society) Papers​

• Benjamin & Gannon, AES Paper 4855 (1998):
  Found that THD below 0.1% is generally inaudible with complex music signals. They used listening tests with trained listeners and various levels of harmonic distortion.
  • Source: AES E-Library

---

2. Floyd Toole / Sean Olive (Harman Research)​

• In studies and books like "Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms", Toole emphasizes that well-designed electronics with THD below 0.1% are unlikely to be audibly distinguishable in blind tests.
  • Book: ISBN 9781138921368

---

3. Douglas Self​

• In "Small Signal Audio Design", Self suggests that THD+N below 0.01% is transparent for audio applications, and that most distortion in that range is dominated by noise.
  • Book: ISBN 9781138850071

---

4. ITU and EBU Recommendations​

• ITU-R BS.1116-3: Specifies conditions for subjective listening tests and implies that extremely low distortion (e.g., <0.1%) is required for transparency under controlled conditions.

---

5. Historical and Industry Practices​

• High-end DACs and amplifiers routinely aim for THD+N below 0.001% as a standard of "inaudibility", even if that level is often overkill. This is often more about measurement excellence than audible benefit.

---

Summary Table:​

THD+N Level	Audibility Summary	Sources
>1%	Clearly audible in most content	AES, general consensus
~0.1%	Possibly audible in test tones, not in music	AES Paper 4855, Toole, Self
<0.03%	Generally inaudible in music	Douglas Self, Harman studies
<0.01%	Transparent and overkill for most listeners	Industry standard, Hi-Fi tests

I find it hard to believe that they could separate inherent speaker distortion from amp like that, at even well above 1%...heck they may actually prefer it.

 

[PristineSound]

I find it hard to believe that they could separate inherent speaker distortion from amp like that, at even well above 1%...heck they may actually prefer it.

The folks at Schiit seems to agree with you.

 

[watchnerd]

This always becomes a long discussion. Other than transducers, it just isn't an issue except with broken designs. Low noise, low distortion is a dime a dozen.

Kind of funny it became more of a concern once it really is not one.

I saw a graph for a OCL tube headphone amp with 2H at about -60 dB

 

[Balle Clorin]

SINAD is the least important parameter for music reproduction.
Think about it we listen to vinyl with a SINAD at 60dB at best without any problems...
At a listening level of 85db and a domestic background noise of 30dba at best .. The difference between -80db SINAD and .-120db is just not audible at the listening chair

I have a a tube headphone amplifier with a 100hz hum 66db below my listening level. It is dead silent when the music stops, cannot hear it, still quiet if I turn up the volume

.maybe it is only there when i measure??

 

[Purité Audio]

Vinyl actually does have a few problems.
Keith

 

[PristineSound]

SINAD is the least important parameter for music reproduction.
Think about it we listen to vinyl with a SINAD at 60dB at best without any problems...
At a listening level of 85db and a domestic background noise of 30dba at best .. The difference between -80db SINAD and .-120db is just not audible at the listening chair

I have a a tube headphone amplifier with a 100hz hum 66db below my listening level. It is dead silent when the music stops, cannot hear it, still quiet if I turn up the volume

I don't know why you would use vinyl's lack of performance as a barometer.

Distortions and noise on vinyl is audible.

 

[Balle Clorin]

Vinyl is just an example of how bad thing can be and still sound OK. I am fully aware of the inferiority of vinyl

Vinyl vs streaming , Note a live recording..not much noise or distortions audible here, the main disturbance the neighbours TV downstairs,,and me fiddling with the iphone

Vinyl first 2 minutes - then streaming 2minutes- and so vinyl again

noisy vinyl?

 

[PristineSound]

Vinyl is just an example of how bad thing can be and still sound OK. I am fully aware of the inferiority of vinyl

Vinyl vs streaming , Note a live recording..not much noise or distortions audible here, the main disturbance the neighbours TV downstairs,,and me fiddling with the iphone

Vinyl first 2 minutes - then streaming 2minutes- and so vinyl again

noisy vinyl?

I don't think anything recorded on YouTube can be used for any sorts of experiment or comparison.

The quality of the mic and recording method, the room acoustics being recorded, the compression used by YT. Then the playback. Ted Denny of Synergistic Research is known to do this type of nonsense to "demonstrate" his products.

But I would go by the scientific papers that I reference on what THD+N is audible.

I personally think the first priority in power amplifiers is being frequency and load invariant on frequency response, then THD+N, because it can compound as you go through the signal chain.

I still look for low THD+N, but I don't think you need more than two zeros, even though I own the Benchmark AHB2.

 

[Balle Clorin]

It is no problem to judge noise level from Youtube since click and pops are really audible... even with iphone in room recording..
But I know adding soundtrack is better like I did here, a quite horrible measuring, and noisy tube amp using a jittery Chromecast analog out with comprimised frequency response, It cannot get much worse than this , but can you hear it? can you hear the hum at -60db??

People do not hear 0.1% distortion at all, even 1% is not objectionable. Stereophile has some test cd with different% distortion, and you could make your own in Adobe Audition etc. anything below 0.2% is hard to hear, and Vinyl has 0.1 to 10% without causing any concern, and speakers can be worse...

 

[PristineSound]

It is no problem to judge noise level from Youtube since click and pops are really audible... even with iphone in room recording..
But I know adding soundtrack is better like I did here, a quite horrible measuring, and noisy tube amp using a jittery Chromecast analog out with comprimised frequency response, It cannot get much worse than this , but can you hear it? can you hear the hum at -60db??

It is just absolutely ludicrous to suggest any type of sound over YouTube or any recorded means for evaluation, experiment or comparison. I find this idea so wildly offensive to human intelligence, that to even talk about it is giving it attention that it does not deserve. In fact, I am very disappointed that Erin Hardison over at Erin's Audio Corner started to do this crap on his YouTube reviews, absolutely did damage to his reputation from my view. I will not discuss this topic any further as I do not want to be dirtied by it.

Absolutely absurdity to the 9th degree.

People do not hear 0.1% distortion at all, even 1% is not objectionable.

Maybe people of a certain age and with certain amount of age related hearing loss. In fact, I've read on multiple literature that age related hearing loss can begin in the mid 20's or earlier. Most people don't hear upper frequencies for the most part of their adult life. I'm pretty sure I can't hear the difference at .1%, but I am pretty sure that those who are trained and in their early twenties or younger can hear .1%.

But the point is if you have .1% in your DAC, then your preamp adds another .1% and your amp adds .1% and then your speaker adds 3%. Then that is a problem. Start off with .001% and add minimally incrementally, it's not a bad idea.

Most people won't have an issue with THD+N if you have competently designed equipment. I don't believe there is an audible need to get anything beyond .00X% unless you want to.

 

[Balle Clorin]

You are missing the whole point. If you do not know what 0.01 0,1 1 or 3% distortion sound like you cannot make a proper judgment about what matters
Even if youtube compression degrades the sound, it does not conceal obvious hum and distortions.


You can learn a lot by actually measuring and listing, ang add distortion to recordings or run music trough an amp with known distortion, then you will be amazed how much deterioration you can you can tolerate. anything below 0.2% on a single tone is really hard to hear and on music it impossible to notice. I know I have tried it
(It you add one distortion stage with a given 2 order distortion it can actually be cancelled by the next stage, if it is in opposite phase. I have measured that in my tube amps too.)

Do you know what compression you tube does to the sound or are you just guessing, I do have high resolution files and recordings uploaded via youtube and the original files to compare?. the degradation is there but sligh. There is no problem hearing differences in cartridges for instance , easy to verify by extracing the audio file too, I have done it-I can even differentiate between turntables/tonearms from you tube extracted audio.

The youtube files cut off the top end nobody hear anyway, and some compression is going but there is no reason to be "catholic "about it the main character and impression remains- provided you stay away from in-room-recordings by a iphone., and listening on your mobile phone. You should try to add a CD audio track to a Youtube and then download it an compare before you are so seriously offended.

I think you are to dismissive and pessimistic of what can be heard,and measured,,even on youtube..

If you can not hear -60db hum and 0.5% distiortion and some db drop at 10khz in the tube amp track I posted it is not youtube fidelity that is the problem ,it is you ears, and mine too, I cannot hear the hum , distortion or loss of high frequency with that track.

Unraveling the Mystery: Is YouTube Audio Really 128 kbps? - SoftHandTech

As the world's largest video-sharing platform, YouTube has become an integral part of our daily lives. With billions of hours of content available, it's no

softhandtech.com

'

 

[PristineSound]

You are missing the whole point. If you do not know what 0.01 0,1 1 or 3% distortion sound like you cannot make a proper judgment about what matters
Even if youtube compression degrades the sound, it does not conceal obvious hum and distortions.


You can learn a lot by actually measuring and listing, ang add distortion to recordings or run music trough an amp with known distortion, then you will be amazed how much deterioration you can you can tolerate. anything below 0.2% on a single tone is really hard to hear and on music it impossible to notice. I know I have tried it
(It you add one distortion stage with a given 2 order distortion it can actually be cancelled by the next stage, if it is in opposite phase. I have measured that in my tube amps too.)

Just because you can't hear .1%, doesn't mean someone who has listening training and in their early 20's with a perfect hearing can't. I probably can't hear the difference myself at .1%, but my point is:
(1) I refer to scientific literatures and go by that, such as the reference I posted earlier, and if new scientific studies comes out then update with that
(2) THD+N can compound throughout the signal chain, maybe you are starting off with .1% but it is not inconceivable that you can end up at 1% by the time it gets to your speakers.
(3) SINAD chasing has become a thing with people and on ASR, I think .00x% is likely sufficient, I don't think from an audibility standpoint, you need <=.000x%.

Do you know what compression you tube does to the sound or are you just guessing, I do have high resolution files and recordings uploaded via youtube and the original files to compare?. the degradation is there but sligh. There is no problem hearing differences in cartridges for instance , easy to verify by extracing the audio file too, I have done it-I can even differentiate between turntables/tonearms from you tube extracted audio.

The youtube files cut off the top end nobody hear anyway, and some compression is going but there is no reason to be "catholic "about it the main character and impression remains- provided you stay away from in-room-recordings by a iphone., and listening on your mobile phone. You should try to add a CD audio track to a Youtube and then download it an compare before you are so seriously offended.

I think you are to dismissive and pessimistic of what can be heard,and measured,,even on youtube..

Maybe you should start a business where you buy a pair of $35k KEF Blade and record different tracks on it, then stream them on YouTube so that anyone can enjoy a $35k pair speakers on their smartphone?