COOLCAT said:
1)in my field if you can't fully "trust" the data its called Marginal and therefore results are inconclusive.(bad- underdeveloped- questionable-debateable).
2)How do you win the "trust" data?
You get whats called "confidence data".
3)How do you obtain "confidence data"?
Measurements.
Luckily, for the Semiconductor industry, voltages, current, and resistance is agreed upon by the global community and can be measured by agreed upon tools/meters!
This doesn't appear to be the case with "sound" measuring devices.
"gets you half way there", it appears:
I don't agree. I don't think so, anyway.
Ethan is obviously way ahead of me, but here's my feelings on the use of say a basic tool like an RTA. This is of course with a nod to the idea that none of this really tells you how things sound, as far as good or bad, and that flat isn't the goal, really.
It seems the interpretation is where many people fall down. They mistake their interpretation as "trust data". They base their interpretations on incomplete confidence data.
If you set up speakers in a room, set up a mic and RTA, and take a pink noise response reading, that's hard data. A very small amount, and almost useless by itself, but hard data nonetheless. That would seem to me to be the beginning of collecting a set of confidence data, not the end. A more complete set would be many measurements in different places.
Still, these mean nothing, really. More is needed, namely the response of the speakers by themselves. I don't mean looking at the stupid charts you get in the box. I mean taking them outside, away from reflections, up on stands, and measuring and listening to them. Say they sounded honky when they were in your room when you walked off-axis. Do they sound that way outside?
And for more confidence, map out your room on the ground, place the speakers inside where you plan to have them, and make a bunch of measurements and listen to them as if they were inside it, and so were you. Then do your inside measurements in the same spots as you did outside.
Now you have two sets of data, measurements taken from the same places, in a free space and in your room. I'd say you can be fairly confident that the differences in the data will be from the room's effects on the speaker's response.
All confidence data so far, if I understand the concept. And yes, I've done it, with all my PA speakers and my home monitors. With the PA speakers, I was looking to establish basic eq and crossover curves as a baseline. With the monitors, I just wanted to see what they were doing by themselves.
More sophisticated gear could give you most of this data much faster. I'm just trying to show how you can gain a higher level of confidence with even modest gear. But even with more sophisticated tools your "trust" data still needs to come from asking yourself questions and experimenting. Say you have a dip in response at 1k. From your confidence data, you have established it is happening in the room. How do you find it or fix it? How can you tell if it's due to a reflection from your console or table, or because your speakers are the perfect distance from the walls to affect 1k, or if it's bouncing off the ceiling?
Your trust data comes from investigating these possibilities to the point where you can say you trust that the dip in response has been traced, after gathering confidence data about each of the possible causes.
Seeing a cancellation dip and slapping some auralex on the walls is not using "trust data". It's making an interpretation based on incomplete data. At least that's what it seems like to me.
That's my personal beef with auto-eq, it can't gather enough "confidence data" to be effective, especially with regard to dips in response caused by cancellations. They do not respond to eq. I've seen more than one guy fry his subs by pushing the 63hz all the way up, not realizing that having his subs on a 2 1/2 foot stage was cancelling most of it out front. Nothing against it, anyone who says a system sounds better afterwards should probably be using it, because obviously they need help, but again, unless the person has walked around the whole listening area, and determined that indeed the change has helped overall, it's low-confidence data.
Wanna have some fun? Set up a Driverack, hook it to a speaker with no tweeter, and hit auto-eq. It will have a heart attack. Or move the mic around while it measures. Or put it down your pants. Good fun.