Discussion in 'Sound Science' started by jthew, Dec 4, 2017.

1. Here's a new 'Sound Science' question for y'all:

Take 4 tone generators tuned to 200, 400, 800, and 1600hz. Turn each one on individually, and set it's volume to 80dB. Now turn on all4 at the same time. What will be the measured dB?

When I change the volume of any of the 10 bands on my AVR, the total perceived volume changes. So I'm guessing it's some integral of the individual frequencies?

2. If 80 dB for the individual tones is RMS value then all four RMS level will be 86 dB, peak level depends on the phase differencies. If the phase differences are all zero, the peak level will be about 91 dB.

So you have 10 octave band equalizer? I suppose it's music/movies sound now instead of 4 tones? Yeah, tinkering with any frequency band should affect the perceived loudness as long as the sound contains the frequencies tinkered.

3. OK. Thanks. Yes, there is a 9 band eq built into the software on the AVR.

Now, I understand that I should make cuts instead of boosts when applying eq, and that having the higher frequencies much louder than lower frequencies could damage tweeters. But let's stick to theoretical for now. Start with some classic rock playing at 80 dB. Using parametric eq, put everything flat to 2kHz. Do a straight line ramp from 2k to 3kHz so that 3kHz is up 15 dB. Next, do a straight line ramp from 3k to 4k such that 4kHz is 25dB above the baseline. Next, stay at +25dB from 4 kHz to 10kHz. Finally, bump everything above 10 kHz to 0.

What would be the measured dB?

4. That's the general rule, because boosting just 3 dB means pushing double power at that frequency! Boost can be used, if you know what you are doing. Cuts are safe.

Well, not exactly. Tweeters can handle the highest frequencies well. It's the lower treble frequencies that are the problem. The best way to destroy your tweeters is to use underpowered amp and overdrive it. The strong bass frequencies (where most energy is in music) will be clipped creating STRONG higher frequency distortion harmonics which goes to the tweeters and burn them up. Using powerful amp which doesn't clip is safer for the tweeters, the energy stays at lower frequencies.

Music with higher frequencies much louder than lower frequencies hardly sounds pleasing to anyone. Music has spectrum that goes down with frequency.

+25 dB at 4 kHz sounds crazy as the ear is very sensitive at that area. Should sound pretty piercing! The measured dB depends on the spectrum of the music (and it changes in time). Maybe you'd get something like 90 dB in average? Impossible to give exact estimates.

sonitus mirus likes this.
5. Thanks, again. I'm trying to make an audiophile hearing aid to take care of my age related high frequency hearing loss. Using the inverse of my hearing test results and my AVR, I was able to make things sound like they did when I was much younger.

The first hurdle is to make sure that I won't do damage to the hearing I have left. The next hurdles will be to find amplification and headphones that can handle the relative high frequency boost. This is the thread I started: https://www.head-fi.org/threads/please-help-design-heavily-eqd-system.866735/ It didn't get very far.

ev13wt likes this.
6. power shouldn't be an issue, if you have a half decent amp and matching headphone it should in any case be able to reach 120db in the 5-15Khz zone?

7. Contributor
all right, here are a few points from both topics, but I have a hard time getting if you're tackling a complex matter with basic approach because you have most of it figured out, or if it's because you're more lost than you think ^_^. so don't mind if I'm being captain obvious from time to time.

1/ are you clear about the EQ you should apply and what you're trying to do in general? the graph given by an audiologist is usually showing a result compensated for equal loudness. meaning that if you have about average hearing, then the graph will end up close to flat.
but if you're at home and use test tones to judge by ear if some frequency is getting louder, you will not get the same graph. I'm not sure if you are already aware of this or not, so just in case: https://en.wikipedia.org/wiki/Equal-loudness_contour

2/ you have an issue turning a graph into an actual correction EQ. that is very natural. you have at least 2 issues:
2A/ a headphone isn't flat. so what can be fairly simple to do on your speakers, is much more complicated on headphone. you can of course decide to apply the same correction and say "F it! I'm leaving the headphone the way it is". but that could leave some massive variations that may or may not add up to your already massive EQ. so it's not easy to tell you if you'll be satisfied, and it's even less easy to tell you if you might finish ruining some frequencies in your ears with some involuntary extreme boosts.

2B/ if you apply EQ on multiple bands, there is of course the possibility that in some places the curves will interact with each other.

the colored areas are individual EQ applied, and the white line is the actual correction I can expect. depending on settings, they can match exactly, or be very different. where I placed my cursor you see in the box the settings I used and I asked for a 4.4dB boost near 2kHz. but the white line is far from it, you can read the actual boost I'll have on the top right, "graph: 0.13dB". that is the result from that boost plus the surrounding EQ settings.

3/ is the matter of having a device and a headphone able to apply a good enough EQ. by good enough I mean with more than kiddo settings so that you can decide how high and how wide some bands should be set. not all EQs will offer something like a 30dB boost(or attenuation). with portable gears you'll often get 6 to 10dB max(10dB feels twice as loud).
then however you plan to set your EQ, let's say you want a boost of 20dB at 5kHz, in the end you'll need to have 5kHz as is, and reduce the rest by 20dB(to avoid clipping the signal). so you'll end up asking your amplifier (or amp section of the portable device) to go in this example, 20dB louder than your usual listening level. it could be a challenge for some sources and some headphones.
that concern was brought up in the other topic. it's a double issue. first you need a headphone that is able to do your preferred loudness plus 20dB or whatever it is that you wished to apply. and next is the amplifier, that needs to be able to give you your normal loudness+the 20dB to compensate the EQ attenuation.
you might not actually need a full 20dB depending on the headphone's signature and part of your EQ, but it's fair to start with that much in your estimate when looking for the right gears.
so you do need a fairly sensitive headphone, one that already has a bright signature would be a good idea. an even better idea would be to see a max SPL value given in the specs(when you'll blow up the headphone of melt the coil). they don't always provide that. I'm guessing if you contact a brand and explain your situation they will give you the number, but you never know.
if you're a quiet listener this might never come to be an issue, but again it's always better to know where we stand.
and of course you need a source that will be able to push your sound at your desired loudness+the value of your max EQ. we can get a fair estimate of that for amplifiers, but it might be less easy for some portable players where specs are often scarce or incomplete enough to be useless.

3/ is it bad for you?
in the end it's obvious that by making the frequencies you can hardly hear way louder to get them fine again, you're also likely to accelerate the speed at which you might completely lose those frequencies. all being a matter of how loud and how much time you spend listening that way each day. but that's the issue with any hearing aid. you decide to get them because the situation without them is too bad anyway. so instead of the idea of ruining your ears, it becomes the possibility to enjoy sounds a little longer thanks to the aids.
doesn't mean you should go crazy, and in general I would still advise to try and get the supervision of some medical professional. but if I was in your situation, I would probably be doing exactly what you're planning to do.

none of the stuff I mentioned are impossible and even really hard to improve or check at a practical level( well, getting the real flat EQ for your ears is something all audiophiles struggle with on headphones). but the sum of them could make for quite the work in research, learning, and testing mostly.

8. If you level your peak out at 95 dB then the music level will be 70 dB.

Sounds safe to me if you don't listen all day.

9. Contributor
sure, it's really all a matter or how much compensation will be applied, where might matter a little for the headphone, and of course how loud the user will actually want his sound. here as we're dealing with a situation of bad hearing at some frequencies, it's safe to expect normal listening level for the non boosted frequencies, and normal listening level+boost for the rest. for me and my quiet listening habits, it's still a piece of cake. but for Amirm and his expected 120dB peaks, it's another story ^_^.

10. Could you please explain a little more?

11. All 4 tones would result in 86dB RMS, but the peak would be 88dB, not 91dB, and would occur every 2.5ms.
This is a widely held fallacy. Clipping amps don't destroy tweeters, excessive power destroys tweeters, which makes the reverse case more true: if you want to burn out tweeters get a more powerful amp. If you look at the spectral distribution of music you'll see that there is far less power in the upper frequencies than in bass. In fact, most of the power is in bass. If you analyze the spectral distribution of clipped music you see that the basic power vs frequency distribution does not change, even with unlistenable clipping because the harmonics created by clipping are below the music content. Yes, they are horribly audible, but there is no significant additional tweeter-burning high-frequency content caused by gross clipping.

Darren G likes this.
12. Boosts up to 10dB are probably safe in most cases. Here's why:
While it is true that a gain change of 3dB means doubling the power, the real power that increases as a result of a gain EQ setting is a function of the distribution of energy, the bandwidth (Q) of the EQ section, and the statistical energy distribution over time. As a result, a single fractional octave bandwidth Eq adjustment of 10dB will be very unlikely to require 10dB more of amplifier power unless a sound exactly at the EQ section center frequency occurs at a very high level, at or near to amplifier clipping. If you analyze the peak energy in a single octave band you find it at least -3dB to as much as -10dB below the total spectrum. A 1/3 octave will be much lower. So in the end you can boost probably up to 9 or 10dB with a relatively narrow EQ and not be in danger of clipping.

Darren G likes this.
13. The peak to RMS ratio of music is highly variable from 20dB to as little as 6dB or less. Depends on the music and how much aggressive peak limiting has been applied. For design purposes a 20dB ratio is assumed max.

14. Thanks for the correction pinnahertz. You're correct. It's 88 dB.

How do you get excessive power to tweeters? Blasting music at high levels with a powerful amp? No, because at high frequency the power peaks tend to be short impulse like sounds. However when you clip long bass notes etc. with an weak amp, you generate long treble content for the tweeter and the voice coil might burn. speakers can handle quite a lot of short peak as long as the rms power stays low. A "100 W" speaker can typically handle short peaks of 500-1000 W if the music is dynamic.

15. Nope. Wrong. Here's the spectrum of a 200Hz sine wave driven 10dB past hard clipping. Note the level of the harmonics. I used a sine wave so you can easily see what kind of harmonics are generated. I used 200Hz because it's an area of spectrum where lots of bass occurs. Note the level of the first harmonic in tweeter range (past the crossover).

Yes, and that makes my point even more true.

sonitus mirus likes this.