Understanding RMS

[Bump Music]

I might be simplifying this a little too much but:

If two tracks have the same RMS levels, does that mean they will sound just as "loud" as one another?

Cuz in my experience that is not the case.

Thanks!

[dementedchord]

in a somewhat generallized sense yes they are as loud..... stop the hissing in the cheap seats..... what you are probably describing is the fact that the energy is not distributed in the same freqs....

[cerberus]

most rms measurements seem to measure right and left channels. to me this implies that the additive effect of right plus left might be better accounted for by measuring the rms of mid and side. perhaps it might narrow the discrepancies between subjective perceptions of loudness and objective rms measurements.

jeff dinces

[dcollins]

Originally Posted by cerberus

most rms measurements seem to measure right and left channels. to me this implies that the additive effect of right plus left might be better accounted for by measuring the rms of mid and side. perhaps it might narrow the discrepancies

No.

What would reduce the discrepancies would be for people to understand that RMS and perceived level have nothing to do with one another........

Play a 500 Hz tone into your fancy "RMS" reading meter. Set it to a comfortable monitor level.

Now switch to 5kHz.

Reads the same doesn't it?

Sounds what? Louder? Softer? The same?

You ran from the room?

Why is this so misunderstood?

DC

[Bump Music]

Originally Posted by dcollins

No.

What would reduce the discrepancies would be for people to understand that RMS and perceived level have nothing to do with one another........

Play a 500 Hz tone into your fancy "RMS" reading meter. Set it to a comfortable monitor level.

Now switch to 5kHz.

Reads the same doesn't it?

Sounds what? Louder? Softer? The same?

You ran from the room?

Why is this so misunderstood?

DC

I don't even need to run this experiment because I now know exactly what you're talking about. Makes perfect sense. Donno why I didn't get it before.thumbsup

[cerberus]

dave, i don't think that just because you are right, it means that i am wrong: i was assuming the same mix mastered by 2 engineers, both keeping the frequency profile as close to the mix as they can, but handling dynamics with different goals in mind. it is possible that the one with lower rms could be perceived as louder. i don't think in that case we will find a peak at 5k on one but not the other.

you have answered the question for bump... but i think i can prove that i am also correct: try inverting the polarity of either the r or l channel of a mix or master. this would not change its rms measurement. however, the inversion would have a huge effect on perceived loudness when both channels are played in stereo. perhaps the bass might go away entirely. i think we don't need to run this experiment either to realize it. thus i'd assume that if one measured the rms of mid-side instead of right-left, then cancellations and re-enforcements between the right and left stereo channels would be more likely to show up in the numbers.

jeff dinces

[minister]

problem is 'LOUDNESS' is a highly personal, subjective impression based on psych-acoustic and physiological factors. ones auditory acuity, inabilities, the environement as well as learned preferences all come into play. even individuals can have different listening opinions at different times.

there have been plenty of studies that have established this.

a VU meter only approximates momentary level changes in program material. its response does not correspond to the overall impression of the program loudness that we perceive. furthermore, both the PPM and VU meters also have a flat frequency response over the entire audio spectrum and therefore do not address the fact that the human auditory system is non-linear with respect to frequency.

calculation of RMS still does not get us all the way there,

[structuredloud]

Speaking of meters having more to do with heat dissipation and nothing to do with our fletcher-munsony ears does anyone know of a meter that would enable a reasonable real-life loudness comparison? If not anyone think it would be useful to try and make one? I could take a shot at it in pluggo format.

[minister]

Originally Posted by structuredloud

Speaking of meters having more to do with heat dissipation and nothing to do with our fletcher-munsony ears does anyone know of a meter that would enable a reasonable real-life loudness comparison? If not anyone think it would be useful to try and make one? I could take a shot at it in pluggo format.

DOLBY and a few other organizations are doing just that. the LEQ-A standard is a starting point. DOLBY's "LOUDNESS" meters represent a beginning at trying to do this for broadcast and film purposes. it uses a proprietary algorithm developed over many years. it is not the answer yet...but the LM100 does a remarkable job of distinguishing vocal contour.

how can you 'take a shot at it' in pluggo if you don't have any research to back you up?

[structuredloud]

Originally Posted by minister

a proprietary algorithm developed over many years.

I would be doing just that!

Actually my friend, I imagine I'd start with a large-ish fft window, examine the difference between a two critical octaves, let's say 120hz and 3khz, lather, rinse, repeat.

Not saying I have a good chance at challenging Dolby's authority on the subject, (maybe I'm blinded by science) but aren't you ever curious why people can't seem to pin down something that seems so simple with years of research?

There's usueally a not so easy answer to that, but why not get your hands dirty anyway.

[structuredloud]

Back in the 1960's, Emil Torick, Bronwyn Jones, and colleagues at CBS Laboratories developed a loudness meter modeled after human hearing. Simply put, the meter divides the audio into seven bands, weights the gain of each band to match the so-called equal loudness curve of the ear, averages each band with a given time constant, sums the averages, then averages the total again with a time constant about 13 times longer than the first. The resulting signal is applied to a display with an instantaneous response. Okay, maybe not so simple, but the CBS meter was found to agree with listeners within 2 dB (although listeners disagreed among themselves by as much as 4 dB when judging the loudness of a given piece of audio).

This was one of the first truly successful attempts to measure loudness accurately and helped solve the problem. Unfortunately CBS Laboratories does not exist anymore and there are very few CBS loudness meters in existence today, so the use of this measurement technique has become rare.

from http://www.tvtech.com/features/audio...f-TC-dtv.shtml

Sounds like a very expensive circuit, but a very cheap plug-in.

[minister]

sure i am curious. just like a lot of people. and dolby ain't the only one...would be bad if they were. the more looking at this, perhaps the more potentional there is for a decent solution, or unending quackery.

"turn it up, dude! this ROCKS!" or "turn that NOISE down!"

prima facia straight forward, eh?

so...good luck.

do you plan to tackle both instaneous and infinite time domains? by integrating the instaneous peak history of short-term 'loudness' into an average? a single loudness value?

..that's interesting about CBS...

[bob katz]

Originally Posted by structuredloud

from http://www.tvtech.com/features/audio...f-TC-dtv.shtml

Sounds like a very expensive circuit, but a very cheap plug-in.

The CBS circuit would be closer, for sure, but it does not take into account frequency and temporal masking, nor the fact that the human ear judges loudness not by 1/3 octave critical bands but rather in much smaller bands called "barcs" (I hope I spelled it correctly, the back of my mind thinks "barqs" but anyway). Understanding these critical bands are the basis of such developments as MP3 and would be the basis for a much more accurate loudness meter. If anyone knows where the loudness-determining bodies are buried, it's Jim Johnston, creditable as the inventor of perceptual coding. But to be totally fair, Dolby has one of the finest researchers on its staff, Jim Fielder, also a pioneer in this field.

[minister]

Originally Posted by bob katz

If anyone knows where the loudness-determining bodies are buried, it's Jim Johnston, creditable as the inventor of perceptual coding. But to be totally fair, Dolby has one of the finest researchers on its staff, Jim Fielder, also a pioneer in this field.

from what i know, Jeffrey Riedmiller of Dolby is no slouch either. he has been tackling this issue for a long time and can speak very plainly about it. in a conference call with me and several other post engineers, he explained a lot issues very plainly. if you read what he has written you find he has a great grasp on the issues and where further work needs to be done. he does not pretend that we have arrived at a solution yet.

i wonder who else is doing this kind of research? sure some university is conducting either loudness or the relationship of 'metering' (visual display) to 'percieving' (by which i mean hearing + cognintion ... beyond pre-cognitive experiential having).

here's a link to handy, nifty little app called 'AudioLeak' which will do faster than realtime LEQ(a) measurements.

http://www.versiontracker.com/dyn/moreinfo/macosx/29084

[bob katz]

Originally Posted by minister

i wonder who else is doing this kind of research? sure some university is conducting either loudness or the relationship of 'metering' (visual display) to 'percieving' (by which i mean hearing + cognintion ... beyond pre-cognitive experiential having).

Thanks for the interesting link, Minister! What is your real name? I belong to an AES loudness and metering subcommittee but I'm mostly a lurker. A lot of active participants.

[minister]

Originally Posted by bob katz

Thanks for the interesting link, Minister! What is your real name? I belong to an AES loudness and metering subcommittee but I'm mostly a lurker. A lot of active participants.

hi bob. i am not hiding. i guess i just didn't put my name on my sig or in my profile... you can find me at the link in my sig. but, my name is tom hambleton.

yer part of an AES subcommitte? look up Preprint # 6348. edit : also #5900

ciao

[minister]

Originally Posted by bob katz

The CBS circuit would be closer, for sure, but it does not take into account frequency and temporal masking, nor the fact that the human ear judges loudness not by 1/3 octave critical bands but rather in much smaller bands called "barcs" (I hope I spelled it correctly, the back of my mind thinks "barqs" but anyway).

...you must mean 'Bark' which is derived from BarkHausen, Georg Heinrich. a 'bark', AFAIK, measures critical band rate, or a unit of 'perceptual' frequency. you use its scale to convert the physical range or spectrum into a psycological spectrum.

high frequencies are perceived at the middle ear while low freqs at the far end of the basilar membrane. and though this knowledge, you can manipulate the 'audio data' to peceptual code a file because the ear discriminates between the energy in the bad and the energy outside the band.

[bob katz]

Originally Posted by minister

hi bob. i am not hiding. i guess i just didn't put my name on my sig or in my profile... you can find me at the link in my sig. but, my name is tom hambleton.

yer part of an AES subcommitte? look up Preprint # 6348. edit : also #5900

ciao

Thanks, Tom! I'm not a fan of pseudonyms in general, it's just one more step to learn whose behind the post.