Originally Posted by craig john
I'm not "in love" with Audyssey. It's a tool. I use it like a tool. I know what it does well, and I know what it doesn't do well.
Thank you for the detailed response Craig. I now see the source of the confusion. The situation is not as you see it. Please see below.
I can't speak to your level of technical expertise. Nor can I speak to your willingness to try to get the tool to do its' job optimally.
Audyssey is aimed at the mass market. The thesis of the original post was that it is easier to use than an EQ. This means setting up the mic, running the automated system and expecting it to work. If you are doing more than this, then you are adding complexity that takes away from the very notion of this system and how it is used by its buyers. But this is not important in the context of this response to you as I the reason for disagreement is a misunderstanding of the data and underlying causes. Please see below.
I can however speak to my own experiences with Audyssey. Here are some examples of the successes I've had with it: My own system: Bass Response, (subs only w 80 Hz crossover):
I can't see these graphs. But I do see the attachments. All of them, in addition to the results you post of others show the bass response to 250 Hz. Those are improper measurements to use in this conversation. I have explaining two problems with Audyssey:
1. Midrange dip. This is happening in the region of 1 to 4 Khz. Clearly this is not visible in a frequency response which only extends to 250 Hz.
2. The slope of the target curve. For obvious reasons, this graph must span the full frequency response from 20 Hz to 20 Khz. Again, stopping at 250 Hz gives us no idea at all what the achieved target response is.
What you do show is the benefit of an Auto EQ system in taming resonances (peaks) that occur in low frequencies. I already explained that this is a benefit of Audyssey as it is of just about any other Auto EQ system. Neither I, nor Harman has criticized this aspect of the system. So not sure why you would post a dozen measurements, all showing how you have smoothed the response in bass region. This is to be expected and is good.
The bad is that some baggage came with it in the form of #1
. Please see more below.
ALL of these systems sounded BETTER post Audyssey, and every one of these system owners PREFERRED the post-Audyssey sound. In fact, they were all thrilled with the improvement in sound quality. I have never had anyone not prefer the post-Audyssey sound, and I have probably measured and tweaked 25 or more systems.
You say you don't know what my skill set is. I am going to turn that around and say I don't know what critical your listening abilities are either. What I do know is my own and that of Harman trained listeners. We all seem to be adept at hearing frequency response anomalies. To wit, I have taken their double blind speaker tests and in both cases, my outcome was the same as their trained listeners.
The word “blind” is key here. Since you don’t state otherwise, I am assuming all of your tests were sighted. It is abundantly easy to convince yourself you are doing some good when performing sighted tests when in reality you are not. No, just because the differences are “big,” it does not mean placebo takes a vacation. Time for a little story
I was teaching my two sons how EQ works. I would set up a filter and then with a checkbox in the UI, I could turn it on and off to show the effect. I did this a couple of times but on the next try, when I asked my children if they could hear the difference as I toggled the checkbox, they gave me that reluctant nod of “we can’t hear it but we are too polite/afraid to say otherwise so we are going to pretend you are right.”
Noticing that, I look at the UI and I see that when I thought I had the checkbox on, it was really off! So all the improvements I thought I was hearing was not there at all! Of course subsequent to this knowledge, I could now hear that it had made no difference.
So unless you can come back and show how your testing has been blind, I am afraid I can’t put much weight on testing which included a huge fan of the technology no less. I can get 100 cable believers to all say they are hearing night and day difference in two identical cables. That kind of data point doesn’t counter the results of blind tests where such bias could not be introduced.
Importantly these testers need to be exposed to something better than Audyssey. Only then do they have a proper reference of what great sound is.
This is why I question the skill level of the person who ran the Audyssey calibration/EQ for the Harman comparison test. I'm not the least bit surprised that nobody liked or preferred the Audyssey in that test. It's a terrible result. But it's clearly not representative of the potential results that can be acheived with Audyssey.
You say the Harman results are “terrible.” I don’t know how you can say that when everything their research shows is embedded in the design of Audyssey. I updated the graph to make this clear:
The Harman test has two parts: 1) a double blind listening test and 2) objective measurements trying to correlate system performance with the listening tests. The latter does not lie. It is what the system is doing. And that is shown in the above graph.
On the left of the graph I have annotated the two problems with Audyssey and drawn an arrow to its frequency response which is the teal colored graph at the bottom. The dip is there as shown by the circle. And the flat response is my eyeball average shown in that flat line.
Now here is the key thing: both of these are by design
. The midrange compensation as they call it is there as they say it is. And the flat response is there if you look at the target curve it attempts to achieve. Given this, there is no way that outcome is wrong. Audyssey is doing what it is supposed to and as such, there is no operator error. Let me repeat: Audyssey is doing what it is supposed to. The tests are 100% valid as far as operating the system.
So the worry then turns to what the listeners heard. There, the test is not what you think it is. They did not ask listeners, “here is Audyssey on and now off; which is better?” They played all of those systems in a random manner using a computer. The trained listener then judged in his mind which sounded more real to them. The no EQ test was just one of the stimulus unbenounced to the listeners. Ratings were then assigned to each system by the listeners, again not knowing which is which. The no EQ system garnered certain score that was below the best performing systems since we know the room interactions deter from performance of the system and by fixing them properly, we can improve the sound (Harman proves this in another research paper). This was demonstrated with two Harman implementations (#1
) and one third-party (#3).
Audyssey and Anthem ARC as a whole garnered scores that were lower than straight wire control. They just did. In the opinion of *every* listener in that test, they overall experience was inferior to doing nothing. You can’t dispute this outcome. They scored without knowing which is which and in balance, they thought using these two systems made the experience worse.
Detailed results shows strong disdain for the midrange performance and bass. We have talked at length why these two occurred and therefore, there should be no objection to this outcome. If you screw up the midrange of a speaker then people are going to cry foul. Likewise, bass is very important to listeners with respect to fidelity. Make it anemic (subjectively) and listener preference goes down.
Yes, the peaks were taken out of the low frequencies. Notice how the Audyssey line in Teal is more flat than dashed line which was no EQ. This, it did. Just like you have seen in the measurements of your system and that of other vocal fans of Audyssey. But with it as I noted, came some baggage such as keeping a rather flat response and introducing a dip in mid frequencies which proved offensive to listeners.
Now, it should be said that critical listeners are much more ruthless when it comes to audio artifacts. I am trained in audio compression and even slightest amount of it can make it unbearable to me to listen to it. Take US DBS radio: XM. It has so much compression artifacts that I can’t listen to it even though I like the content they have. With some 25 million users, clearly this is not nearly as much of a problem for them. So maybe if Harman had you or the other 25 people you mentioned that their scores would be higher. That doesn’t change the objective facts that Audyssey introduces errors in the system. The errors exist and we have found them fair and square. We use trained listeners so that we can find and fix these problems and not stick our head in the sand by having non-critical listeners tell us it is all fine.
Is Audyssey going to be as good, or as flexible or adjustable as your $20,000 JBL system? Obviously not. In fact, if your system isn't significantly better, you spent a LOT of money for nothing. But that doesn't mean that Audyssey is "worse than no EQ." It's only worse than no EQ if it's setup badly, and not optimized.
I hope you now agree that the outcome was correct and Audyssey did what it was designed to do. There is a reason that in so many years, not one study has been published to dispute its results. There are things that are wrong with the system and the research found them subjectively and objectively.
As to comparison to JBL, that is not in discussion. What is in discussion is use of low cost EQ systems. There, I can outperform Audyssey easily with just a bit of work and possibly much less than you are doing Audyssey . Everyone can learn to do the same and get the bonus of really understanding room acoustics.
Nothing is more educational than pulling a peak down in bass frequencies and turning that on and off and listening to the results. Or changing the overall frequency response. No amount of screwing around with Audyssey gives you that. Not even close.
If Audyssey was doing its job then sure, I am as lazy as the next guy and would use it.
But it just doesn’t. As such let’s not give advice to people to ditch an EQ and get an AVR with Audyssey.