How loud is loud, how to measure it? Is THX calibration bad for your health?

[sarumbear]

By litterature? Notches are common in the 3-6 kHz region.

https://digscholarship.unco.edu/cgi/viewcontent.cgi?article=1132&context=capstones

Notches are common doesn’t it doesn’t happen in lower frequencies. The study you linked is about sounds generated by acoustic instruments and specifically at high frequencies. There’s no study done on amplified sound at low frequencies on that study.

 

[Thomas_A]

They studied sounds generated by acoustic instruments and specifically at high frequencies. There’s no research done on amplified sound at low frequencies on that study.

I was referring to ear damage and that notches are mostly found in the high frequency range. (Also, industrial workers have notches in high frequency spectrum.) But perhaps you misunderstood my previous statement ("established that ear damage due to high SPL is mostly found in the high frequency range")?

And, if it was known or very uncertain that permanent ear damage is caused by high SPL low frequency notes, I am quite certain that the study with the 120 dBSPL constant 30 Hz tone fed into human volunteers would have been ethically rejected.

 

[sarumbear]

I was referring to ear damage and that notches are mostly found in the high frequency range. (Also, industrial workers have notches in high frequency spectrum.) But perhaps you misunderstood my previous statement ("established that ear damage due to high SPL is mostly found in the high frequency range")?

And, if it was known or very uncertain that permanent ear damage is caused by high SPL low frequency notes, I am quite certain that the study with the 120 dBSPL constant 30 Hz tone fed into human volunteers would have been ethically rejected.

We haven’t seen any frequency or level on any research so far. Your reference does not answer the issue either: does low frequencies are safer than high frequencies. So far you are only assuming. Will you risk your ears with wishful thinking?

 

[EEE272]

Notches are common doesn’t it doesn’t happen in lower frequencies. The study you linked is about sounds generated by acoustic instruments and specifically at high frequencies. There’s no study done on amplified sound at low frequencies on that study.

I did like the article because you candraw some conclusions - of course, very careful and I don't say it is proof - the highest exposure is percussions with >130dBC. Nevertheless, they only get to ~90dBA. The difference is probably caused by an extreme contribution in the low end.

All instruments are around 90dBA but maxed out at ~120dBC. That is 10 dB difference to percussions.
Yet, the hearing loss of the percussions is not as bad as for, e.g., woodwinds.

Thus, it seems true that low frequencies are harmful, but probably not as much as the higher ones. The question is by how much you can increase them...

 

[sarumbear]

The question is by how much you can increase them...

Indeed!

 

[Thomas_A]

We haven’t seen any frequency or level on any research so far. Your reference does not answer the issue either: does low frequencies are safer than high frequencies. So far you are only assuming. Will you risk your ears with wishful thinking?

There are research among musicians saying that high SPL from music instruments can cause ear damage. How frequent is that in musical instruments with primarily low bass notes (double bass, cello)? From the literature I cannot say since I don't have access to all:

"Results showed that 21.7% of the 3,292 musicians responding perceived having a problem with hearing. Findings showed the highest rate of occurrence was in rock/alternative musicians; in musicians who were included in the nonclassical grouping; and in musicians who played amplified instruments, drum-set, and primary brass instruments."

Instrument-specific Reports of Hearing Loss: Differences between ...: Ingenta Connect

www.ingentaconnect.com

And as I referred to earlier; a study with a static 120 dBSPL 30 Hz signal fed into the ears would not be accepted ethically, if the signal was changed to 4 kHz. Music where the 30-40 Hz dynamic signal average 120 dBSPL would be very high in volume, if it was the typical music spectral content of a dynamic music file. It would be very high also if it was the peak level. A 85 dB average SPL with a dynamic music, as I measured in my room, is of no concern IMO as long as the music is naturally dynamic and played occasionally. Heavily compressed music can be a problem though.

 

[Thomas_A]

Here was another one measuring the SPL at musicians ears in orchestra.

Sound Exposure of Symphony Orchestra Musicians

Abstract. Background: Assessment of sound exposure by noise dosimetry can be challenging especially when measuring the exposure of classical orchestra musi

academic.oup.com

 

[Soundmixer]

So if 80 is your setting (-25 on your calibrated receiver), then -25dB is the highest you can produce (105dB-25dB=80dB).

This is not correct. The headroom allocated in the cinema digital system is 20dbs, not 25dbs. At 80dba, the maximum volume would be 100dbs, not 105db.

 

[Chromatischism]

And as I referred to earlier; a study with a static 120 dBSPL 30 Hz signal fed into the ears would not be accepted ethically, if the signal was changed to 4 kHz.

My WAG, probably because we get natural amplification from our ears at that frequency:

Among other natural phenomenon.

 

[EEE272]

This is not correct. The headroom allocated in the cinema digital system is 20dbs, not 25dbs. At 80dba, the maximum volume would be 100dbs, not 105db.

That is strange. THX standard says that -20dB pink noise should play at 85dbC.
Thus, if you lift it by 20dB, you get 105 dBC. Now, I indeed do think that you want some headroom to not clip the signal, so it would maybe be 103 dBC? Isn't that related to the crest factor, which can be smaller than 5dB.

In any case, a signal should be able to reach close to 105dBC.


Edit: Oh wait, is it because of A weighting? Is pink noise 80dBA equal to 85dBC?
I assumed the question was about 80 dBC but I might have missed that it was 80dba.

Edit 2: in any case, the question was ultimately about something entirely different anyway.

 

[Soundmixer]

That is strange. THX standard says that -20dB pink noise should play at 85dbC.
Thus, if you lift it by 20dB, you get 105 dBC.

This is what you said previously, and what I was responding to.

"So if 80 is your setting (-25 on your calibrated receiver), then -25dB is the highest you can produce (105dB-25dB=80dB).

I was addressing the contradiction of your two statements. -20 is correct. -25 is not. Gut check here. Nobody uses 85db as a reference level anymore. Most content mixed and mastered for streaming and physical disc uses 77-79db as a mastering level. That means no higher than 97-99db which is much more tolerable (but still very loud) than 105db. I would refer to Audio2920 for information on the average sound level he uses for cinema-based mixing.

 

[EEE272]

This is what you said previously, and what I was responding to.

"So if 80 is your setting (-25 on your calibrated receiver), then -25dB is the highest you can produce (105dB-25dB=80dB).

I was addressing the contradiction of your two statements. -20 is correct. -25 is not. Gut check here. Nobody uses 85db as a reference level anymore. Most content mixed and mastered for streaming and physical disc uses 77-79db as a mastering level. That means no higher than 97-99db which is much more tolerable (but still very loud) than 105db. I would refer to Audio2920 for information on the average sound level he uses for cinema-based mixing.

Oh, no that was a total misunderstanding. Sorry about that! I was answering to the question of a user who said he/she obtains 80dB on a calibrated system. He/she is not a mixer.

Edit: that is why I said there was already some confusion around the question itself. I believe, ultimately, it was about whether you can at all obtain sound that is louder than 105. Which you can if you set your amplifier higher than 0 dB.

 

[witwald]

Edit: Oh wait, is it because of A weighting? Is pink noise 80dBA equal to 85dBC?

For comparison, I believe the C-weighting is approximately –3dB at 30Hz, while the A-weighting is approximately –40dB at 30Hz. I'm not sure how the pink noise energy integrates across the nominal 20Hz to 20kHz frequency range to produce a dBSPLA or dBSPLC reading. The dBSPLC reading would be the higher of the two, possibly by around 6dB or so.

 

[Andysu]

THX high frequency screen boost.

 

[audio2920]

Now, I indeed do think that you want some headroom to not clip the signal, so it would maybe be 103 dBC? Isn't that related to the crest factor, which can be smaller than 5dB.

Not totally sure where you're going with this, but as you say; No, you can't turn it up 20dB without clipping because the crest factor is greater than 0.

Incidentally, clipping is sort of allowed in cinema (it's only frowned upon, not disallowed) so crest factor can go pretty low on some sounds, and in the analog domain you can exceed +20dBu slightly because of the overshoot.

In home ent mixes, because the limiters are normally "true peak" they act on an interpolated peak value (as well as having finite attack/release/look ahead of course) so there's virtually no additional headroom needed above the -2dBTP (or whatever the limit is set at)

 

[EEE272]

I have one more question about loudness - which might open a can of worms...

The dialnorm is half a mystery to me. If I understood it correctly, it should roughly adjust the dialog to be at -31 dbFS. For broadcast, this makes sense and it seems to be used also for home DVD/Bluray.
Nevertheless, how does this mess with the reference loudness (as far as there is one - let's assume someone would really like to follow the rules).

I noticed that Netflix seems to use the dialnorm to adjust movies to the -27 LUFS that they target.
Usually, the dialnorm is less defined compared to this Netflix approach, as it seems to just state: measure your dialog and make sure it is -31dbA - where the definition of how to extract it is not entirely defined - but please correct me if I am wrong.

Now let's come to an example.
In case of Transformers Age of Extinction, the dialnorm shows up as a -8dB in the sound information on Netflix.
Nevertheless, the same movie on Amazon Prime has a dialnorm of +4dB. A whooping 12 dB difference.
Still, when you measure with an SPL meter, the difference between the 2 is merely 6dBA.
Now, it could be that we have two differently mixed versions - maybe one home and one cinema?
Btw. the +4dB Prime sounds like my Bluray that I have of the movie.
I know that Atmos seems to not have a dialnorm but I would be curious to hear about the situation with (for example) DD5.1.

What would be your guess, what source we have on Netflix and Prime?
Is there any educated guess to be made as to what level this was likely mixed at?
For example, in case of Netflix, we know that after dialnorm we have -27LUFS, so we know that if the dialnorm is -8, then the "original" was at -21 LUFS. Given some kind of average loudness that productions have (and that reference levels are probably between 75-85), can we derive anything from this?
I know we cannot be certain, but it feels like it might be possible to recognize at least some ballpark number.
Would someone have an idea?

 

[Andysu]

I have one more question about loudness - which might open a can of worms...

The dialnorm is half a mystery to me. If I understood it correctly, it should roughly adjust the dialog to be at -31 dbFS. For broadcast, this makes sense and it seems to be used also for home DVD/Bluray.
Nevertheless, how does this mess with the reference loudness (as far as there is one - let's assume someone would really like to follow the rules).

I noticed that Netflix seems to use the dialnorm to adjust movies to the -27 LUFS that they target.
Usually, the dialnorm is less defined compared to this Netflix approach, as it seems to just state: measure your dialog and make sure it is -31dbA - where the definition of how to extract it is not entirely defined - but please correct me if I am wrong.

Now let's come to an example.
In case of Transformers Age of Extinction, the dialnorm shows up as a -8dB in the sound information on Netflix.
Nevertheless, the same movie on Amazon Prime has a dialnorm of +4dB. A whooping 12 dB difference.
Still, when you measure with an SPL meter, the difference between the 2 is merely 6dBA.
Now, it could be that we have two differently mixed versions - maybe one home and one cinema?
Btw. the +4dB Prime sounds like my Bluray that I have of the movie.
I know that Atmos seems to not have a dialnorm but I would be curious to hear about the situation with (for example) DD5.1.

What would be your guess, what source we have on Netflix and Prime?
Is there any educated guess to be made as to what level this was likely mixed at?
For example, in case of Netflix, we know that after dialnorm we have -27LUFS, so we know that if the dialnorm is -8, then the "original" was at -21 LUFS. Given some kind of average loudness that productions have (and that reference levels are probably between 75-85), can we derive anything from this?
I know we cannot be certain, but it feels like it might be possible to recognize at least some ballpark number.
Would someone have an idea?

It's exciting I noticed same, Thing.
Start a thread on list your movies at dialnorm.
Also you can switch it off on the AVR or is it, sounds like I can hear difference when it's witched off a level increase.

I have to set up the laserdisc that shows.
Some discs I see mentioned before.

The Black Hole. Dialnorm +4dB

 

[audio2920]

there any educated guess to be made as to what level this was likely mixed at?

It's a good thought, but I don't think it's possible to conclude much about the original level based on dial norm and/or known platform loudness specs I'm afraid. The setting of this parameter is not entirely arbitrary of course, but as you've discovered, implementation is highly inconsistent.

Anyway, if I *had* to guess, Amazon have just left the value at default, without relationship to the conent as it is probably pre-normalised to their spec. Netflix could have the cinema version as -8 is quite a lot of reduction were it to be a mix already formatted for home ent. Although it could almost as easily be that it's home ent that's somehow been made way outside their spec!

Acquisitions are often treated differently to original content, so the VOD platform may take whatever has been made already, rather than necessarily having a new mix made to their spec.

I can't see any particular correlation between the numbers that leads to anything conclusive, but maybe I'm not looking hard enough

 

[Andysu]

Ain't Laserdisc great.

 

[EEE272]

Acquisitions are often treated differently to original content, so the VOD platform may take whatever has been made already, rather than necessarily having a new mix made to their spec.

I am sure they are lazy in this case but they do run the normalization on whatever content they used.
BTW. are you able to hear if a content is likely mixed for 79 or 85? What are indicators to listen for?

Not by ear, but analysis, I realized that most blu-rays limit the LFE to around -15dBfs. The BEQ catalogue shows the maximal signal for the low-end frequencies for many uhd and blu-ray discs.
Thus, is it likely that in these cases, someone mastered everything 10 dB lower than for cinema? Or is it common in a cinema mix to stay below -10dB in the LFE?