Frequency Analysis Window Decibel Scale - wrong ? (er, no...)

Report · Jan 28, 2022

i'm using 2017 Audition to try to analyze energy distribution of music so i can design speakers that properly allocate available power to different frequency bands ( woofers vs midrange vs tweeters ) ...

i am having an issue ...

Audition on most tracks shows 30 decibel differential between peak of bass kick at 50 hz and begin of high frequency rolloff at 10 khz ...

that is to say it shows 50 hz as -20 db and 10 khz as -50 db ...

screen shot below is analysis of the entire track " PSY - Gangnam Style " for example ... but other tracks look similar ...

but over at a speaker building forum i saw data posted that shows only 15 decibels for the same drop:

https://techtalk.parts-express.com/forum/tech-talk-forum/64801-musical-genre-specific-average-spectr...

15 decibels is suspiciously HALF of 30 decibels, yet the overall shape of the curve is actually basically the same ...

they both show peak at 50 hz and rolloff beginning at 10 khz ... but the SCALE on one of them seems to be off by a factor of 2X ...

it's also a little suspicious to me that bass peaks in Adobe are already at -20 db, which is kinda low ?

when i design the speaker closely following popular professional designs its power capacity versus frequency ends up matching the curves posted on partsexpress tech talk, which show 15 decibel drop ... which leads me to suspect Adobe scale is measuring something else, or displaying it wrong ...

one possible reason for 2X factor is Voltage vs Power ... 2X Voltage = 4X Power ... though when correctly implemented both will result in 6 decibels but, perhaps it is NOT correctly implemented ?

is the frequency analysis panel any different in more recent versions of Audition ?

even one extra decibel in sound reinforcement can cost thousands and we're talking about 15 decibels of discrepancy here ...

i don't need to know where the error is coming from necessarily just which is the right figure.

thanks in advance.

Report · Jan 28, 2022

Audition cannot measure power. The scaling is all voltage-based (same as SPL) and has been correct for many years. Nor has it altered. I suspect that you are simply confusing the units. The power, as expressed in dB would indeed come out at half of the SPL values measured.

Report · Jan 28, 2022

Thank you for responding Steve !

That was my original thought, that because doubling of voltage results in quadrupling of power that this may be the reason why one scale is off by 2X relative to the other ...

But i Googled this yesterday and 2X Voltage should be 6 decibles and 4X Power should also be 6 decibels ...

In other words there should be no difference whether Voltage or Power is measured - decibels should be the same ?

Report · Jan 28, 2022

also from Wikipedia:

http://en.wikipedia.org/wiki/Power,_root-power,_and_field_quantities

"A root-power quantity is a quantity such as voltage, current, sound pressure, electric field strength, speed, or charge density, the square of which, in linear systems, is proportional to power.[3] The term root-power quantity refers to the square root that relates these quantities to power."

Report · Jan 28, 2022

No - as in what your last answer states: "That is, a change in power by a factor of 10 corresponds to a 10 dB change in level. When expressing root-power quantities, a change in amplitude by a factor of 10 corresponds to a 20 dB change in level."

As I said, Audition cannot measure power, and as a consequence displays voltage, which is the same as what an SPL meter would read if put in front of a loudspeaker. If you could measure the power (which believe me is rather harder to do) then the equivalent reading would be halved.

If you really want to understand this well, then you have to look into how this works in acoustic terms.

Report · Jan 28, 2022

Ok i figured it out. The answer was in the original post on PartsExpress Tech talk forum the whole time - of course i haven't actually read that post until about 2 hours ago when i finally got desperate enough to do it.

In his post he mentions "I did check that everything was working properly by looking at Sheffield Labs Uncorrelated Pink Noise 20Hz to 20kHz and it is flat"

As soon as i read that i immediately ran pink noise through Audition and here is what i got :

this isn't flat. this is 10 db / decade drop off.

Wikipedia to the rescue once again !

https://upload.wikimedia.org/wikipedia/commons/7/77/Pink_noise_spectrum.svg

It also shows 10 db / decade drop off, just like Audition ...

now let's read the description from Wiki:

http://en.wikipedia.org/wiki/Pink_noise

"In pink noise, there is equal energy per octave of frequency."

https://en.wikipedia.org/wiki/Colors_of_noise

"The spectral power density, compared with white noise, decreases by 3.01 dB per octave (density proportional to 1/f ). For this reason, pink noise is often called "1/f noise"."

so actually ... the spectral power density of pink noise falls off at 10 db / decade ( as shown by Audition ) ...

but at the same time energy per octave is constant ... as show by REW ( Room EQ Wizard ) which the poster on PartsExpress Tech Talk has used to come up with his data ...

so i was wrong when i assumed one program scales the results by factor of 2X versus another ...

it's actually that Audition has an additional 10db/oct slope compared to REW due to different ways they represent the same thing ... this adds just over 20 decibels of additional drop from 50 hz to 10 khz which is why where REW shows about 15 decibel drop from 50 hz to 10 khz Audition shows about 35 db ...

i guess the moral of the story here is that there is a reason why things like pink noise exist - without it i would be screwed !

thanks everybody for suffering along with me through this mess !

Report · Jan 28, 2022

from Wikipedia:

http://en.wikipedia.org/wiki/Decibel

"Two principal types of scaling of the decibel are in common use. When expressing a power ratio, it is defined as ten times the logarithm in base 10.[5] That is, a change in power by a factor of 10 corresponds to a 10 dB change in level. When expressing root-power quantities, a change in amplitude by a factor of 10 corresponds to a 20 dB change in level. The decibel scales differ by a factor of two, so that the related power and root-power levels change by the same value in linear systems, where power is proportional to the square of amplitude."