How quiet is -60dB?

[617]

Well, that makes sense. I do like the idea for relating it to the extremes that people think are important. So many people are sure they can hear ~-140 dB (like, whether 24-bit audio is plain-truncated or TPDF-truncate). It can be frustrating trying to get across just how tiny that is.

Yeah for some reason people have a sort-of reasonable idea of the upper frequency limit they can hear, but the dynamic range that 24 bits can theoretically encode is pretty mind boggling.

 

[Blumlein 88]

I've had a couple threads on this and seen others over the years. Some with test tracks and polls etc. The results are pretty consistent.

People listen seriously at average levels of 70-80 db SPL. There are peaks of 15-20 db above that. If you introduce artifacts at increasingly lower levels people stop hearing them somewhere around -60-70 db or usually around the 10 db SPL mark. While there are of course exceptions and some artifacts are more noticeable than others in our hearing's most sensitive range this outlines the envelope.

Yes rooms have 35-45 db SPL ambient noise, but most of that is low in frequency and in the range of 3-5 khz many rooms have 10 db SPL or a bit less. We can hear into noise a little bit so none of this is surprising.

All of which points to 96 db of CD being both sometimes needed and all that is required like 98% of the time (yes made up statistic, and also probably correct). If you throw in a a little dither you can drop the low end a bit more meaning the point at which 16 bit isn't enough is less than 1% of the time.

As a self check, you have LP and cassette with dolby. LP has maybe 60 db to work with. It too has lower levels in our most sensitive range. And even on the quietest discs we can hear a little noise in near dead quiet portions. Most of the time music masks it enough we aren't aware of it and it has not a huge impact on what we hear. Cassette with dolby fits into this range too. If you have dolby C, you've got more range than LP and it does sound quieter. As a further check I found 7.5 inch per second pre-recorded reel to reel inadequate. The level of hiss is too high. Most of the time it is masked by the music, but it takes very little for it to be heard. You can learn to ignore it for your pleasure, but it is easily heard. Here the level of hiss is 45 or 50 db.

 

[Kvalsvoll]

I made test sound samples some time ago, they are useful to find what -60dB is like:

https://www.kvalsvoll.com/Articles/TestSignals/Loudness 880hz -20dB sweep -60dB.flac

Here, a 880Hz tone is at -20dB is mixed with as tone sweep at -60dB. So the sweep in the background (which clearly is audible, even at fairly low volume) is -40dB below the 880Hz tone.

Here it is -60dB below that same tone - now the level of the sweep is -80dB, and now it is just barely audible at low volume:
https://www.kvalsvoll.com/Articles/TestSignals/Loudness 880hz -20dB sweep -80dB.flac

I have samples at -100dB and -120db as well, the -100dB is barely audible if you turn up the volume (0dB):
https://www.kvalsvoll.com/Articles/TestSignals/Loudness tone sweep -100dB.flac

All those samples can be found in the
https://www.kvalsvoll.com/Articles/TestSignals/
catalog.

Now you can also find the samples with tone sweep only, and then you can hear the -120dB, it is just a matter of turning up the volume to dangerous levels.

Note that the next sample in the list below -120dB tone sweep is a 0dBFS sweep, just mentioning, in case.. I think you get the idea of a potentially rather unpleasant scenario.

 

[Frank Dernie]

These comparisons to distance don’t really work for me. They suppose that there is no noise at all in the environment and also that it’s free space. Neither of these are a reality and therefore I cannot really imagine hearing a worker 18 miles away. Note that your average home noise is already around 40dB. Distortion products needs to first get louder than this noise to be able to hear them.

They work for me simply because they illustrate the numbers and what they really mean.
Yes the background noise level isn't 0dB but that doesn't matter in getting a feel for the magnitude of the numbers and the audibility.
Imagine going outside and trying to hear somebody 18 miles away.
Then imagine still being able to hear that if you were listening to music at a normal level.
What it means is that the numbers we are often quoted as audible or required minimum performance are way way more than we could ever hear in reality rather than on paper.
IMO.

Here's an experiment you can try. In a quiet environment it's not unusual to score -60 dB or lower. And that understates what we can actually hear. This experiment only tests your short-term dynamic range acuity. Long-term dynamic range is wider because the ear uses biological compression mechanisms to compensate.

Dynamic range acuity is a different thing to distortion audibility and this beautifully illustrates how unlikely it is for us to hear distortion at a level were we can't even hear any sound at all in a quiet room!

 

[Frank Dernie]

After all, I can listen to music at a moderate level, turn it down 60 dB, and still make out every word of the lyrics without effort, even in the midst of music.

Can you?
If I do the same I can't hear anything, never mind make out the lyrics.
I usually listen to good dynamic classical music recordings with my level control set 20dB below maximum power (with compressed rock music it has to be -30dB or lower to have a similar average loudness FWIW).
If I turn the level down to -80dB, ie 60dB less, I have to walk to the speaker to hear anything at all and hear nothing at my listening seat.
It is easy for me to test since I have a volume control calibrated in dB.

Edit:
To make myself clear I am meaning the loudness of distortion products relative to the music, not absolute level.
If I play back a typically compressed modern rock track, where the level is pretty well all in the top 20dB of the level meter at -60dB it is all very intelligible but at -80dB not.

 

[BrokenEnglishGuy]

I made test sound samples some time ago, they are useful to find what -60dB is like:

https://www.kvalsvoll.com/Articles/TestSignals/Loudness 880hz -20dB sweep -60dB.flac

Here, a 880Hz tone is at -20dB is mixed with as tone sweep at -60dB. So the sweep in the background (which clearly is audible, even at fairly low volume) is -40dB below the 880Hz tone.

Here it is -60dB below that same tone - now the level of the sweep is -80dB, and now it is just barely audible at low volume:
https://www.kvalsvoll.com/Articles/TestSignals/Loudness 880hz -20dB sweep -80dB.flac

I have samples at -100dB and -120db as well, the -100dB is barely audible if you turn up the volume (0dB):
https://www.kvalsvoll.com/Articles/TestSignals/Loudness tone sweep -100dB.flac

All those samples can be found in the
https://www.kvalsvoll.com/Articles/TestSignals/
catalog.

Now you can also find the samples with tone sweep only, and then you can hear the -120dB, it is just a matter of turning up the volume to dangerous levels.

Note that the next sample in the list below -120dB tone sweep is a 0dBFS sweep, just mentioning, in case.. I think you get the idea of a potentially rather unpleasant scenario.

I had to cranck the volume in my headphone ( night test ) for hear the -100dB, if anything was playing i couldn't hear the thing at -100dB because is so tinny, in speakers i think can be more hard because of the distance.

 

[Absolute]

Hear for yourself, download attached wav. One channel has a constant tone (C), other channel has a tone a tritone above it at 0, -10, -20 etc.

https://www.dropbox.com/s/a9i4ajevss3lb7u/0 to minus 60.wav?dl=0

I can still hear it at -60 dB through headphones, albeit very faint. -50 dB is clearly audible in this scenario.

It depends on how far the harmonics are removed from the fundamental and at which frequency it is (masking).
Tones and music are different things.
A S/N ratio of 50dB is audible because the average level of music will be 10-20dB below peak from which S/N ration is calculated.
Noise will thus be 35dB down below average music levels.
It also will depend on how loud the peak/average SPL is.
In music our dynamic range is about 70dB, add 20dB and 90dB S/N ration is fine.

Unless you have gear that can reach 120dB peak and you want absolute silence standing close to the speakers. In that case you may need 120dB S/N ratio because standing close to the speakers you want silence but you listen further away (lowering the noise at the listening position)

Also those that listen for noise 10cm away from their tweeter but listen 2 meters away to music will need a much higher S/N ratio than those that only care about noise at the listening position. The picture of the working men illustrates this well.

When your distortion products are not masked they also need to be low but because these distortion products are narrow in BW and noise is wide distortion can be a bit higher than noise before it becomes audible. For nasty sound you need substantial amounts of distortion (depending on music content as well because this can mask)

Some folks are bothered by noise, others less so. Also noise spectrum matters. It's complex, certainly with music.

Needless to say... It is not extremely costly nor difficult to build electronics with low noise and distortion (high SINAD). For vinyl and mics this is another matter and getting low noise is a bit more demanding/difficult.

Thank you for the detailed response.
This stuff is bothering me because I don't find numbers to correlate good enough with my experience. Noise especially.

I can clearly hear hiss from my speakers at 3,5 meter even though I can't reliably measure it above the room noise from that distance. I don't have the LP measurement here, but going close to the speaker it's very clear that there's alot of noise; (absolute level not calibrated here)

Room noise is around 30-35 dB from 1-8 khz

According to numbers it shouldn't be possible to discern this noise due to random room noise being higher from the listening position, but I guess it's explainable through the specifics of the spectrum of the noise vs room noise and not so much the relative spl in itself. I can still hear it at around 65 dB avg music playing. My guess is that most people react to amplifier noise, not amplifer distortion.

It's about 699 billion miles, about 7500 au.

Wow, Australia has its own distance unit?
It sure makes sense.. G'day mate! Sure, nearest bar is just 2 au that way. A few weeks walkabout I reckon, nothing much!

I've had a couple threads on this and seen others over the years. Some with test tracks and polls etc. The results are pretty consistent.

People listen seriously at average levels of 70-80 db SPL. There are peaks of 15-20 db above that. If you introduce artifacts at increasingly lower levels people stop hearing them somewhere around -60-70 db or usually around the 10 db SPL mark. While there are of course exceptions and some artifacts are more noticeable than others in our hearing's most sensitive range this outlines the envelope.

Yes rooms have 35-45 db SPL ambient noise, but most of that is low in frequency and in the range of 3-5 khz many rooms have 10 db SPL or a bit less. We can hear into noise a little bit so none of this is surprising.

All of which points to 96 db of CD being both sometimes needed and all that is required like 98% of the time (yes made up statistic, and also probably correct). If you throw in a a little dither you can drop the low end a bit more meaning the point at which 16 bit isn't enough is less than 1% of the time.

As a self check, you have LP and cassette with dolby. LP has maybe 60 db to work with. It too has lower levels in our most sensitive range. And even on the quietest discs we can hear a little noise in near dead quiet portions. Most of the time music masks it enough we aren't aware of it and it has not a huge impact on what we hear. Cassette with dolby fits into this range too. If you have dolby C, you've got more range than LP and it does sound quieter. As a further check I found 7.5 inch per second pre-recorded reel to reel inadequate. The level of hiss is too high. Most of the time it is masked by the music, but it takes very little for it to be heard. You can learn to ignore it for your pleasure, but it is easily heard. Here the level of hiss is 45 or 50 db.

This makes more sense than anything. Thanks!

 

[Frank Dernie]

I had to cranck the volume in my headphone ( night test ) for hear the -100dB

Possibly, but the whole point of such a test is to find what levels you can actually hear without turning the volume up from your normal listening level.
This gives an idea of what you would be able to hear in dynamic range when listening to music.

 

[voodooless]

They work for me simply because they illustrate the numbers and what they really mean.

That is the whole point: it really doesn't work for me... Last time I checked, our world is not flat nor noiseless or empty.

 

[Frank Dernie]

That is the whole point: it really doesn't work for me... Last time I checked, our world is not flat nor noiseless or empty.

We are at cross purposes.
They tell me I am how likely to be able to hear the sound of just a distortion product in the presence of a music signal.
It is because the world, at least the music my listening room, is not noiseless that the idea of being able to hear a distortion product equivalent to somebody shouting various distances away whilst listening to music that I found interesting and worth testing on myself at home.
Maybe I just like it because it confirms my experience
It shows the absurdity of some of the claimed performance requirements.

 

[Absolute]

It shows the absurdity of some of the claimed performance requirements.

How does this stuff work in the chain?
120 dB SINAD in a dac is all well and good, but that stuff needs to be amplified by 40x with an amplifier with 100 dB SINAD.
Is the noise then accumulated, like dac noise + amp noise x 40 = audible hiss?

 

[Frank Dernie]

How does this stuff work in the chain?
120 dB SINAD in a dac is all well and good, but that stuff needs to be amplified by 40x with an amplifier with 100 dB SINAD.
Is the noise then accumulated, like dac noise + amp noise x 40 = audible hiss?

I have tested me. For my ears 80dB SINAD is adequate, just, so anything better thgan that is unlikely to sound any difference to me.
I decided to test myself because LPs can sound quite good on my system and they have a SNR of at best -70dB and distortion way worse.
In fact how good an LP sounds depends far more, IME, on the quality of the recording and production of the disc than on the poor measurements of a record playing system, but still, if SINAD even as good as CD were actually required for good SQ then a record player would never sound anything other than awful regardless of recording.

My recommendation to anybody is to check it out for yourself on yourself. It is not difficult.

Then you know whether buying something "better" will be worthwhile to you personally.

 

[RayDunzl]

120 dB SINAD in a dac is all well and good, but that stuff needs to be amplified by 40x with an amplifier with 100 dB SINAD.
Is the noise then accumulated, like dac noise + amp noise x 40 = audible hiss?

Adding the decibels on my nakkin, for those two noise sources (120 DAC and 100 amp) - the result should be around 99.95 for uncorrelated noise from each.

 

[Absolute]

I have tested me. For my ears 80dB SINAD is adequate, just, so anything better thgan that is unlikely to sound any difference to me.
I decided to test myself because LPs can sound quite good on my system and they have a SNR of at best -70dB and distortion way worse.
In fact how good an LP sounds depends far more, IME, on the quality of the recording and production of the disc than on the poor measurements of a record playing system, but still, if SINAD even as good as CD were actually required for good SQ then a record player would never sound anything other than awful regardless of recording.

My recommendation to anybody is to check it out for yourself on yourself. It is not difficult.

Then you know whether buying something "better" will be worthwhile to you personally.

I'm very relaxed on distortion numbers and would say -80 dB distortion in an amp is likely not audible, maybe unless we're talking about a rampant degree of higher order distortion.
According to these simple tests we've seen here, -60 is surprisingly audible even if theory suggests that means I can hear Mr. Shoutie from 1,16 miles away. Which means that this whole concept is useless for our application, but fun nonetheless.

If that 80 db Sinad refers to noise then that's absolutely audible for most people on most speakers after amplification. It's the hiss-problem

 

[Frank Dernie]

If that 80 db Sinad refers to noise then that's absolutely audible for most people who sit inches from their speakers on most speakers after amplification.

FIFY

 

[Absolute]

Adding the decibels on my nakkin, for those two noise sources (120 DAC and 100 amp) - the result should be around 99.95 for uncorrelated noise from each.

That's horrible! Imagine noise and distortion almost 100 dB louder than the signal!

FIFY

Hah, indeed. Or quiet rooms or sensitive tweeters, or both. THat's me. And it's way too audible

 

[Frank Dernie]

Hah, indeed. Or quiet rooms or sensitive tweeters, or both. THat's me. And it's way too audible

If I set my "default" volume setting the voice at full scale is 80dBA at my listening position and I can hear him say -48dB from my listening seat well enough to hear his enunciation of 48. Below that I am aware of perhaps a couple more levels (to -54dB and maybe -60dB) but not loud enough to tell what he is saying.
If I go about 2ft from my speakers I am aware of some sound down to what must be -66dB, counting down, but not clear enough to actually discern his enunciation of 66.
When playing no music but the system on and volume set as above my system produces no audible hiss at all, even with my ear against the tweeter.
This is pretty well what I have found before here, ie from my listening position things go quiet at -60dB (edit, relative to my normal listening level), which is an absolute sound level of 20dBA in here which is around the background noise level.
Obviously If I set the volume to be 10dB louder the lowest sound I can discern goes up 10dB too.
Given that I can just be aware of the -60dB speech in a fairly quiet room it still seems unlikely that I would hear a distortion product of that loudness when listening to music.
No idea what hiss level I would be able to hear if there was some, though, because there isn't any

 

[abdo123]

Speaking of -60 dB, that is roughly the distortion level of the Sennheiser HD 800 S, which is the lowest distortion measured on a headphone in this forum (so far).

I wonder if anyone with a trained ear can hear the distortion on them (between 1k and 5k).

 

[MRC01]

...Dynamic range acuity is a different thing to distortion audibility and this beautifully illustrates how unlikely it is for us to hear distortion at a level were we can't even hear any sound at all in a quiet room!

True, I was replying to the OP, which wasn't about distortion, though this thread has now shifted to that topic.
Typical of test results I've seen, short-term dynamic range is around 60 to 70 dB in a quiet environment, while distortion can be detected at -40 to -50 dB under ideal conditions. So distortion is harder to detect.

... I wonder if anyone with a trained ear can hear the distortion on them [Sennheiser 800S having distortion at -60 dB] (between 1k and 5k).

Perhaps but I doubt it, as -60 dB is lower than any level I've seen distortion detected in actual testing.

 

[BrokenEnglishGuy]

Speaking of -60 dB, that is roughly the distortion level of the Sennheiser HD 800 S, which is the lowest distortion measured on a headphone in this forum (so far).

I wonder if anyone with a trained ear can hear the distortion on them (between 1k and 5k).

Hd800 doesnt have the lowest distortion, even dan clark aeon have much lower distortion, also the hd800 with correction EQ raise a lot the distortion in bass. Amir calculator 35%.