Originally Posted by J_Palmer_Cass
The old "loudness button" did work fairly well when audio was played back at low levels.
Facts: Every manny who included a loudness button had its own version of the EQ curve. The ELC are averages from an unspecified number of people, all of whom have different perceptions of what is equal loudness. Finally, we who get the whole phenomenon of playback of recorded sound understand that how loud 16.5 Hz sine tone has to be to "sound as loud, averaged from 1,000 opinions on the subject" as a 1,000 Hz sine tone at 'x' dB IS IRRELEVANT.
Funny how you doctored the ELC curve that I supplied to suit your own purposes. All the documented curves on that chart differ by 20 dB at 1 kHz. By definition, that is accurate. In the 20 Hz area, all of the curves vary by about 10 dB. That means 5 dB of change at 20 Hz for every 10 dB of change at 20 Hz. The 2 dB calculation that you made up is based on a curve that was doctored by you by intent and is not accurate information.
I doctored nothing. I used the "previous" curves. Previous to what? Why did they need revised? What changed in the criteria to affect such a change?
Let's take your "revised" curves.
10dB at 20 Hz, 4dB at 125 Hz = 7dB average. Cut that in half and you get 3.5dB average in the SW band.
Take, instead, your 5dB suggestion. IMO, if you play music at 0dBRL, then play the same music at -10dBRL with the subs bumped +5dB and you don't hear a FR change, your hearing is too poor to comment in this discussion. Loudness buttons (by any other name) are a detriment. They're certainly not required when listening at -10dBRL with the subs running an average of +6dB hot to begin with (which is the consensus gleaned from this forum over 12 years time) and I wish we could stop reading these errant "because this is how you hear" ELC references.
As far as the SL content following the ELC, that is another misconception made by you. Fine you made an FFT curve fit for the low end. Have you ever looked at the high end on SL to see if it fits the curve? Well it does not do so in any way shape or form. SL is not a "the way that you hear" representation of content. SL is a straight FFT program. You have to use an octave based RTA program in order to "see" how balanced the soundtrack is.
A straight FFT program displays acoustic energy in a fixed bandwidth per bin manner. An RTA program display acoustic energy in a fixed octave per bin manner that is based on how you hear. An RTA adds up all the FFT bins in a fixed bandwidth and displays them in a fixed per octave manner (AKA variable bandwidth).
FFT - bandwidth of one bin at 20 Hz = bandwidth of one bin at 1 kHz (etc)
RTA - bandwidth of 1/6 octave in 20 Hz to 40 Hz (AKA 1 octave) area is about 3.6 Hz = bandwidth of 1/6 octave in the 1 kHz to 2 khz (AKA 1 octave) area is about 170 Hz.
What this means is that the SL FFT program does not display frequencies in a manner that is based on how you hear. With SL, a 1 kHz the signal level displayed on SL will be down 30 dB from the 1 Hz signal level recorded on a soundtrack that is balanced and is based on how you / the mixer hears. With an RTA program, the soundtrack will measure flat with the same content. That is hardly a "smiley curve" that you misrepresented as having a meaning in this application.
Who said SL has anything to do with "the way you hear", and why would we want it to?
SL graphs what is input according to your settings. That's what we want it to do for the purposes of this thread, and it does so quite well.
Here is 10 Hz and 1,000 Hz input at the same level (REW generator only goes down to 10 Hz, but which particular low Hz you prefer is irrelevant):
The input signal was at -12dB for both frequencies. They sure look the same to me by all 4 metrics shown on the SL graph.
For MWB, we're not interested in what is happening in the soundtrack at 1,000 Hz, so we shorten the graphing BW to add resolution below 120 Hz for MWB. When we do that, we get far greater resolution to 1 Hz and guess what? When you input 10 Hz vs 100 Hz, equal input, you get equal output from SL... with greater resolution:
So, yeah, feed SL a Log sweep and it will accurately show the attributes of a log sine sweep. That would be useless information, as we already know this, but the point is that SL will render that log sine sweep as it is input. Input a soundtrack and it will accurately depict that input as well. That would be useful information because every soundtrack is different and we want to see those differences.