9 replies to this topic

[Spectrum]

Why do engineers relish in their 1 to 2 dB compressor and/or EQ tweaks while describing their personalised settings...

...yet a single decibel is typically the absolute bare minimum that us humans can pick out from within our 120dB dynamic hearing range.



Can we really hear the difference between a cowbell dipped 1.5 dB @ 3257Hz compared to the original?

[Captain Terrific]

tis a good question Spectrum...  maybe its because over several tracks, those 1 & 2 dBs sum up to a larger reduction?? i dunno!!

I noticed earlier today that a 2 dB dip @ a certain frequency sorted a problem i was having trying to fit a snare in a mix... 3 dB was too much, lost the lower end punch.

perhaps the 'q' slope has some sort of relevance here? 

[Spectrum]

Perhaps?

Why do engineers relish in the argument re which DAW's bounced/exported/rendered stereo file output sounds best, or that it's even such a delicate process...

...yet, the very same program has had absolutely no problems 'rendering' the exact same audio for playback to the stereo monitors up until that point?

http://bp2.blogger.c...s320/escher.bmp

You know, it's not like the poor ol' video guys and animation artists who must view their work in jerky, low-resolution modes - just to give them the general idea - during the session as that's the best their systems can output due to the enormous CPU-intensive processing that realtime video effects/graphics demands. They must then set their computers to 'render' overnight, long after they've gone home, sometimes for days on end, just to output the final high resolution version so they can see what on earth they'd been working on during the weeks prior.

Meanwhile, us audio guys hear our work in pristine quality as we go.

Except, it all (apparently) goes to shit once the DAW has to stop its realtime playback and instead lay down on the hard drive what was otherwise going to play perfectly from the speakers.

[rhythmboy]

Quote

Why do engineers relish in their 1 to 2 dB compressor and/or EQ tweaks while describing their personalised settings...

...yet a single decibel is typically the absolute bare minimum that us humans can pick out from within our 120dB dynamic hearing range.



Can we really hear the difference between a cowbell dipped 1.5 dB @ 3257Hz compared to the original?

The difference in air pressure between 0dB and 120dB is a factor of a million - 0.00002 N/m2 to 20 N/m2 so a difference of 1db can still be quite significant, especially in the frequency range between 1-4kHz where our hearing is sensitive. An EQ unit may work in a predictable linear or log curve across all frequencies but the ear's response isn't flat. A difference if 1.5dB at 3.7kHz would definitely be noticeable if the original levels were high enough - at 0dBFS a 1.5dB drop is a bit more than 10%. But a 1dB drop at 30Hz would be far less noticeable, especially at low levels to start with.

[ajay]

Im not sure about all the technical lingo (except to agree with RB that 1db differences are certainly noticabel).

Fuck me that cube is seriously messing with my eyes.

[Jester_Fu]

Quote

Why do engineers relish in their 1 to 2 dB compressor and/or EQ tweaks while describing their personalised settings...

...yet a single decibel is typically the absolute bare minimum that us humans can pick out from within our 120dB dynamic hearing range.

Can we really hear the difference between a cowbell dipped 1.5 dB @ 3257Hz compared to the original?

OK - let's get one of your myths straight. Over 70% of the human race can BARELY destinguish a 5-6dB change. No shit. 3dB puts you in the 25th %ile of the human race. Below 3dB and you're talking new born baby grade hearing or wolf like freak of nature. Proper engineers don't buy that shit... we had the argument with one of my old uni lecturers who then made us do a lab with spectrum analysers and freq generators through crossovers we had to built (electronics course... so he was trying to keep it to course material) and then dared anyone in the room to prove the 3dB drop they could see on the SA they could also hear. There were about 2 people who did it. Below 3dB and we all failed.

Where the 1-2dB difference may come into effect is when there are harmonics present and not a pure single frequency signal. If you were to consider harmonics and the effect they have on what we hear and like to hear, then CT's summing theory is of more relevance, IMO, than any of the others. Certainly if you have 3-4 channels all with 1-2dB harmonics over the source frequency then you're going to get 3-8dB increase in intensity over one or more of those harmonics.

[Jester_Fu]

Quote

You know, it's not like the poor ol' video guys and animation artists who must view their work in jerky, low-resolution modes - just to give them the general idea - during the session as that's the best their systems can output due to the enormous CPU-intensive processing that realtime video effects/graphics demands. They must then set their computers to 'render' overnight, long after they've gone home, sometimes for days on end, just to output the final high resolution version so they can see what on earth they'd been working on during the weeks prior.

Myht number 2....

Let's take good old Novo Beef Cheif. The guy runs a small editing business. He does all of his work using Avid and a PC based system. Hee was able to render a full 30-50 minute session in around 2 hours previously. He now has a new HP box allowing him to do it in around 30minutes. You're living in 10 years ago land, speccy.

[dec0n]

Quote

Myht number 2....

Let's take good old Novo Beef Cheif. The guy runs a small editing business. He does all of his work using Avid and a PC based system. Hee was able to render a full 30-50 minute session in around 2 hours previously. He now has a new HP box allowing him to do it in around 30minutes. You're living in 10 years ago land, speccy.

true, well... depends on what you're working with. Alot of the newer HD codecs/hardware are quite efficient, and the PC hardware has definitely caught up bringing render times down etc. But, put a few 3k layers down in Shake or some similar app & render time will still crunch on most large systems.
Same with 3D animators, high res work in Maya or something still needs alot of render time...

[rhythmboy]

Quote

OK - let's get one of your myths straight. Over 70% of the human race can BARELY destinguish a 5-6dB change. No shit. 3dB puts you in the 25th %ile of the human race. Below 3dB and you're talking new born baby grade hearing or wolf like freak of nature. Proper engineers don't buy that shit... we had the argument with one of my old uni lecturers who then made us do a lab with spectrum analysers and freq generators through crossovers we had to built (electronics course... so he was trying to keep it to course material) and then dared anyone in the room to prove the 3dB drop they could see on the SA they could also hear. There were about 2 people who did it. Below 3dB and we all failed.

I'd argue time also plays a big part in this. One of my students a couple of years ago did some research into the 85dB SPL calibration used for monitoring levels when mixing films, asking "why not calibrate music studios the same?". He ran some tests on groups of people, playing them 70-75-80-85-90dB sections and asking them to hear (a) differences at all and ( differences in bottom end specifically.

When the shift in SPL happened suddenly like a switch, people could detect the change quite clearly, over 90% success rate.

But when there was a gap of silence between the SPL changes - even of a few seconds - people's responses plummeted to less than 30% success.

Admittedly the above example used complex signal music mixes, not pure sine waves (which screw up our perceptual accuracy worst of all), and so the argument for cumulative effects of harmonics* and multiple tracks would come into play.

We ascertain absolute SPL levels very poorly. But we can ascertain change in SPL very accurately. However our ability to ascertain how much change is generally terrible. All we know is there was 'a change'.

And I haven't even started on the influence of frequency bands, tones within the ear's critical bands, masking, localisation, reverberation, tinnitus, and so on. The physics texts all seem to point to the fact that hearing response is a largely individual issue, inlfuenced by a lot of simultaneous factors, and any generalisations we make are very very broad indeed.

* In this respect I agree with Jester and CT - comes down to the basics of Fourier's Theorem of analysis and synthesis. If we add two signals of equal frequency, amplitude and phase together, we get a 3dB increase. Increase to 4 signals and we get a 6dB increase. Increase to 8 and we get a 9dB increase. Double the sources = +3dB output. So if there are common frequencies present across multiple tracks, these will sum and create a larger boost than you might expect. Also if you boost a frequency with EQ on one track it adds even more to the cumulative level of all common frequencies in the mix, even those on different tracks.

I also suspect (but need to read to back up) that if a strongly pitched sound with lots of harmonics is assumed, then certain frequencies with a harmonic relationship to each other will have positive and negative peaks occuring at the same periodic time points, and therefore overall level is summed:

Eg 1st harm 10Hz will peak at 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0 secs
Eg 2nd harm 20Hz will peak at 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4 etc
Eg 4th harm 40Hz will peak at 0.025, 0.05, 0.075, 0.1, 0.125, 0.15, 0.175, 0.2, etc

That's assuming the mix is perfectly in phase at all frequencies

[Jester_Fu]

My mixes are always perfectly in phase ;D