Say, when you are at a live (acoustical) concert listening to live instruments, the SPL of the bass is what it is, and the SPL of the high is also what it is. When we hear it we hear a certain ratio of the loudness of the bass vs the high, and that is how they sound "natural" to us. For example, if the SPL of the 1 kHz is 80 dB (= 80 phon) and the 100 Hz is 100 dB (= ~92 phon), the "natural" loudness ratio is 80 phon to 92 phon.
If we listen to its recording at home at a lower volume of 60 dBSPL at1 kHz (= 60 phon), without any loudness compensation, the 100 Hz will be 80 dBSPL (= 62 phon). The loudness ratio is now 60 phon to 62 phon, and is quite different from the original 80 phon to 92 phon. Therefore, we'd want to compensate for it be boosting the low frequencies.
The goal is not to make the 20 Hz sound equally loud as 1 kHz. The music dictates how "loud" the 20 Hz is and how "loud" the 1 kHz is. The mixing and mastering engineers eq the mix to have a ratio that sound "natural" at their specific target listening level. If we listen at a different level and want to reproduce the music with the same loudness ratio, we'll need to apply eq (loudness compensation).
I think it's illuminating and answered the main question. I heard people say that a lot of music is mastered at 75-85db. I'm assuming that's 75-85db a-weighted. I still find the question of how loud I actually listen to music and how loud the speakers have to go to be confusing.
-----
https://www.audiosciencereview.com/...ef-ls50-meta-review-speaker.25574/post-873573 to the post.
The above post in Meta thread is what got me thinking (note the 5 likes). I remeasured with my a weighted db meter and at the loudest I'd ever critically listen using the "max" button (highest spl measured is shown on db meter) with a variety of tracks and got 69.5db, so let's round up to 70db. There's going to be some averaging going on there with the db meter. Subtract 3db from 70 since there will be 2 speakers, another 3 for being close to a wall so 63.5db. Let's say I swap to an OLED tv as a monitor and speakers are now 1m away from me. I have over 23db of headroom for music.
If music is mastered at louder average levels than I listen then I'd be boosting bass a bit so there's going to be some EQ there, but only down to 120hz. I guess the spooky part is if room interferences force me to boost bass frequencies above 120hz and the Metas get stressed.
I don't understand what I'm looking at with their imgur link. I'm assuming they're saying a track playing at 70db on average is somehow peaking at near 97db but I have no idea how or why.
I'm trying to compare the thd graph to the graph of how loud music are at different FR and comparing that to thd audibility graphs and I'm having a hard time. I'm assuming if use a sub and crossed over at 120hz, anything above 120hz in thd would improve over Amir's measurements, or is that only for IMD? I guess I can pull up some spectrograms via Audacity for some of my tracks to see how they land or use this averaged graph from 'AES paper 8960'.
And... this distortion threshold graph I don't entirely understand:
I think distortion is not nearly as bad of a problem as people make it out to be absent of insane levels. Amir listened to Metas and said it's fine unless you want 'the room to shake under you and to really feel the music', while I listen at below average levels. I think distortion fears with Metas are overblown when paired with a sub and listening at sane levels. Still, I'm a paranoid person and I also just like to know a little bit more about the specifics of the limits of what is audible or not, and what could potentially be problems in rare cases, in part just for my own education. So, I list what people criticize about gear I'm considering/already bought and my responses in hopes you or somebody else more knowledgeable can weigh in.
----------------
Dr. Jack Ocklee Brown who helped develop the Metas wrote on ASR in a post you or mdsimon shared about how coaxial speakers are best listened to 10 degrees off axis. Erin said in a Kali IN review that EQ should not be based on on axis response since you're want to EQ based off of at least 10 degrees off axis response, and probably 20 degrees (for the Kali). Do you agree with this assessment? Easy EQ via automatically generated Spinorama github results would probably be off then.
---------------
One area that is a bit unknown with the PA5 is frequency response in to complex loads. Hypex shines in this area as it has frequency response that is very independent of load which is not often the class D amplifiers.
Moving to the amp side, as you probably know pma and Restorer-John often bash cheaper class D amps. Drama aside, pma/Restorer-John has brought up 4 points of contention:
1. Class D amps have problems with FR in 'complex loads'.
2. Effective speaker impedance is not simply the lowest part of the impedance graph. Phase angle matters too. Some amps can be underpowered when 4ohm testing suggests they are not.
3. thd+n behavior of many class D amps are not very linear and vary a lot based on frequency + power. They can also be poor in 15khz-20khz range.
4. Bandwidth for class D amps is insufficient and should be up to 200khz.
Here are my thoughts on each:
1. This is the most concerning one. Simply measuring with a mic and applying EQ? I suppose the problem areas tend to be far above frequencies where the room or desk reflections? affect it? Seems like people can't agree if it's feasible to test FR in 'complex loads' and to decide which load, but then I don't understand how pma managed to do it with just a speaker and a mic. Amir said he tested FR through speaker load and said it was only a problem with cheap <$100 amps.
Seems like FR with the 4ohm (resistive?) load Amir does in his tests is a small indicator of amp behavior. Aiyima the pma tested with the 2db deviation, and in Amir's testing 20khz was off by 0.7db while PA5 was off by 0.2db. So, in this limited jank indicator, PA5 exhibits less than a third of the deviation of the Aiyima. I also can't hear above 15khz, so the problem areas would be 10.5khz-15khz and <1db based on that measurement with Aiyima which I consider worst case. Every measurement in Amir's review except max power output shows PA5 outclassing Aiyima 07 and I don't think I have reason to think it won't be the case here. If the error is <1db I fear measurement inaccuracies with my mic will outweigh the correction and do more harm than good.
Here is pma's testing of Aiyima:
2. I don't think I listen loudly enough for PA5 to struggle. I don't think this matters in my case. pma would bring out EPDR graphs but that only works for class AB amps and we just don't know how class D amps interact.
3. I don't think it's at audible levels, and the power level of music at 15khz-20khz playing normal music rather than test tones is simply too low to matter. I think this issue is a non-issue.
4. I struggle to understand the technical reasons behind this objection, but multiple people who seem pretty technical do not agree there is good evidence such wide bandwidth is important.
Happy New Year to you too! Again, sorry if I'm just bringing up infinite questions that may seem simple. I've done... hundreds of hours of reading around the forum at this point and that's before EQ process even begun. Your patience and knowledge is always very appreciated.
Meta buying time is when PA5 is on the verge of being delivered and that day is fast approaching. Getting some last minute questions in.