Decibels

Continuing discussion from the other thread

Yes, sort of. You guys got it for the most part- it’s a logarithmic scale weighted for different measurements. A-weighted (dBA) is what you see on equipment, and that’s arguably because it yields lower values than the more balanced dBC. A has a more aggressive curve to account for the “human perception,” though it’s usefulness is debatable. For room measurement, you’re better off using C, but for pc fan specs, dBA will give a lower number.

With dBA, the perceived loudness doubles in increments of 10, so… one Bel. decibels are used in favor of bels since dBC’s “doubling” occurs in increments of 3.

Yes, though there are also absolutes for acoustic SPL measurement. 0dB is no energy. Anything above 85dBC sustained will begin to damage your hearing, and the “allowed” exposure time halves with each “doubling” of power. So an hour at 85 is equivalent to 30 mins at 88, and so on.

On a live console, you may notice that each fader is marked in adjustments of 3-5dB near unity, while each increment can be 10, 20, 30 dB, or greater, towards -inf. Oh, and that’s another thing- negative infinity and metering. Historically, audio equipment used electrical levels (dBv, dBm) for metering, then later the nebulous “VU.” Whenever new tech came out, dynamic range increased. Tape, for example, gave a “free boost” in level, which was welcome, considering overmodulating vinyl would break the expensive cutting equipment. As time passed, the units changed to reflect this. Now, we use dBFS for everything digital, where 0dBFS is the maximum possible value for any given bit depth, and all values are negative values in reference to 0, all the way down to negative infinity. Curiously, there is no standardization for dBu (electrical) to dBFS (digital), so bit depths could continue to increase without requiring new nomenclature. Broadcast standards, however, have propelled along commonly accepted levels. This has introduced another entire metering system: LUFS. LU/LKFS is based on decibels, but has some weighting and gating (plus regional regulations) for time-based program measurement. For the most part, you can equate 1dB to 1LU.

As far as bit depths, 24 is *almost all we need for recording, as the available headroom is more than we can use when compared to the actual SPL in a given physical space. In other words, whatever you’re recording won’t exceed the noise floor so far that the tech specs can’t capture it.

*An interesting and welcome equipment trend in the last 10-12ish years has been putting multiple ADCs in each mic input, where the second (inaccessible to the user) is set roughly 20dB lower than the first. The advantage to this is that if the primary reaches 0dBFS, the secondary “safety” track can be cut in. Some gear will even stitch it for you in real time, then save it as either a 24 bit int file with gain reduced (but still optimal SNR), or a 32 float file with no gain adjustment, where the values can exceed 0dBFS. So yeah… there’s some fuzzy math when it comes to floating point, but I’m guessing we’re all familiar with floating point shenanigans. The short version in this context is that nothing can exceed 0 at the DAC, and I have only ever seen one DAC and one ADC that claim to be 32-bit, so everything has to come down to at least 24 int on the way out.

In studio calibration, the reference level for each channel is dependent upon the room size. A large room requires more power from the speaker to project bass than does a small room that struggles to trap the standing waves at the long wavelength of low freqs (also why foam can be pointless if too thin or used improperly). So, in practice, while a theater’s front channel reference levels (blasting shaped pink noise at -20dBFS) are 85dBC for film, a small room only needs 79dBC to feel about the same as the theater. A mix from a calibrated small room will translate relatively well to a large one, but large studios will still spend the money to build a full-sized mix stage- so they can get a perfectly accurate measurement for the room size (and also certifications like Dolby, because we know Hollywood loves its clubs and in-crowds, but that’s a topic for another time). All that said, I find it especially irritating when a theatrical film is painfully loud. They have the range to be, and they mix in a calibrated environment, so every choice is intentional (Snyder’s Superman, grumble grumble…).

To be fair, though, there are no real standards for film. Again, it’s about perceived loudness when it comes to sound. @MCVET your question was a bit open-ended, so I hope that answers it.

9 Likes

This is a terrific insight. Thanks very much!
I’m probably going to need to read this a few more times, to get a proper grasp, and maybe even about some of the related things as well. I have such a poor knowledge of sound that it’s embarrassing, but it’s incredibly interesting to me!
I’m going to let this topic evolve a little bit and then ask about reasonably cheap audio equipment for the home.

This is all really great info, I just want to expand upon something.

This is a good rule of thumb but isnt quite accurate… and I get it, I used to repeat the same things. Frequency matters here.

This video goes over it pretty well

tl;dw our ears have varying sensitivity so damage can occur sooner or later depending on the frequency.

Sorry if I am dragging off topic.

1 Like

Oh, I wanted to point this out. Some audio technicians came to our factory and measured our noise levels. We all had a microphone with a meter to measure the noise levels attached to our shoulders and we averaged at 86 dB. Just under the “Tough working conditions” limit. No pay addition for us :stuck_out_tongue:
But we do have personalized ear plugs. So, that’s something.

1 Like

IIRC it was something like an increase of 3 dB was double the sound intensity, 6 dB double the sound pressure level, and 10 dB was double the sound intensity?


And I believe I linked somebody on the blue forum this a long time ago:

http://www.animations.physics.unsw.edu.au/jw/dB.htm

That is awesome!
Do you know about decibels? I’m sure you do, since you’re an engineer. Would you mind explaining in semi-layman’s terms how decibels scale? I know it’s a logarithmic scale, but that’s all I know :stuck_out_tongue:

It’s pretty simple, it doubles each time.

10dB - > 20dB = twice as loud
10dB - > 30dB = four times as loud

And so on

1 Like

What about 1->2->3->4…and so on.

I’ll leave that to the pro, nonetheless here’s a gain factor chart

2 Likes

Interesting. Thanks!

(patiently waits for the pro to wake up)
Wake up, Mister Freeman. Wake up, and smell the ashes.

Okay I am no sound engineer I am going by memory.

dB are a ratio. That means that you calculate the ratio between the power of two “signals” (sound waves).
The formula for it was something like this -
(have no idea how to start the math mode to write the equation )

10 log (s1 / s2) = x

x - the required ratio
s1 - signal one the one you want to compare
s2 - signal two the one you are comparing to ( reference )
the number in the beginning depends on what exactly you are comparing. From what I remember for sound waves it was 20 not 10 as in my example.

When you see a dB rating on headphones or something usually there is a reference taken for the basic ear. ( I have no idea what that reference is, google is your friend ).

Also

that’s not right but it’s fine you have been working until now.
every time you increase the dB with 10 the difference in the ratio is multiplied by 10.

So 10dB = 1st signal is 10 times stronger than the reference.
20dB = 1st signal is 100 times stronger than the reference.
30dB = 1st signal is 1000 times stronger than the reference.

and that goes on.

2 Likes

Oh, ok. That makes it more understandable!
Thanks for taking the time.

Also:

Lol.

1 Like

X Doubt, I’m speaking of perceived loudness.

But I’m way out of my field, so @khaudio take it from here.

1 Like

Oh wait you are not talking about the ratio itself. Yes I do remember that perceived loudness had some other factors depending on frequencies. Havn’t worked with audio so I am in deep waters here.

This is true. In fact, most hearing loss occurs as a widening ravine centered around 1KHz. Of course, the older you get, the more you lose in the highs, but the reason that remembering 85dBC is a

is because you often encounter broadband noise when out working, not huge narrowband spikes. This is where dBA would be useful. Anyway, yeah, you got it.

1 Like