"Voicing" or "tuning" a speaker after a simulation... cheating?

[fjhuerta]

Quite a provocative thread title, I think.

Here's what I'm thinking. I'm still "fixing" a 3 way design I never truly enjoyed that much - 2 RS270s, 2 W4-1337s, 1 SB-29. The units are overachievers, but I decided on settling on 2nd order curves - and I'm getting all sorts of issues with floor bounce (cancellation), diffraction on the tweeter from, I suppose, driver proximity, and phase not tracking exactly as it should. All this made the speaker sound a bit bright.

And a huge 200L monkey coffin sounding bright is not exactly a good thing.

Anyway, after taking measurements with a lot of care, I found myself with a nice on-axis response, very controlled off-axis curves, and a decent enough impedance.

Yet, I still find the speakers "bright".

And so, I added padding on the tweeter and dropped the level until I found the speakers "good sounding" and with nicely controlled off axis curves. This meant around -1.5 dB output from the rest of the units (when integrated).

Which leads me to think. I cheated. I mean, the speaker should sound fine when designed properly. It's just me and my ideas (that a big speaker should sound, well, bassy and powerful) that led me to alter it via "voicing".

I sort of think most people would think the speaker sounds great when I voiced it as flat as I could.

Does this make sense to anyone else?

[fastbike1]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

My work entails probablistic risk assessment of complex systems. Here's a quote from one of my people, when adding perspective on using modeled results. "No matter how complex your model, the real world is more complex still."

You're listening in a real room with real ears and modifying to your preferences. The modeling program doesn't really model any of the room effects or your listening preferences.

I just don't see why you would think that the simulation program would account for room interaction. You can change the perceived response of the speakers by moving them within the room. Where does the simulation account for that? You also seem to be warring with two compteing mental images: flat response is desired vs "big speakers should be bassy and powerful".

Further the folks that produce their own designs invariably talk about voicing the final crossover to get what they want. Paul Carmody has even mentioned being surprised on the occasion the first modeled crossover sounded "right".

I don't think you're cheating, I think you're completing the design. Everyone's preferences are different. I know that there are entire lines of well regarded commercial speakers that I don't like at all, as in none of several models. I have also listened to some very expensive models that I did not find to be particularly exceptional.

On the other hand, there are a few commercial models that I have really liked. Two in particular are quite different from each other. Some, preference rules.

[Pete Schumacher]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

My work entails probablistic risk assessment of complex systems. Here's a quote from one of my people, when added perspective on using modeled results. "No matter how complex your model, the real world is more complex still."

You're listening in a real room with real ears and modifying to your preferences. The modeling program doesn't really model any of the room effects or your listening preferences.

I just don't see why you would think that the simulation program would account for room interaction. You can change the perceived response of the speakers by moving them within the room. Where does the simulation account for that? You also seem to be warring with two compteing mental images: flat response is desired vs "big speakers should be bassy and powerful".

Further the folks that produce their own designs invariably talk about voicing the final crossover to get what they want. Paul Carmody has even mentioned being surprised on the occasion the first modeled crossover sounded "right".

I don't think you're cheating, I think you're completing the design. Everyone's preferences are different. I know that there are entire lines of well regarded commercial speakers that I don't like at all, as in none of several models. I have also listened to some very expensive models that I did not find to be particularly exceptional.

On the other hand, there are a few commercial models that I have really liked. Two in particular are quite different from each other. Some, preference rules.

+1

[Deward Hastings]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

Here's a quote from one of my people, when added perspective on using modeled results. "No matter how complex your model, the real world is more complex still."

+1

[mzisserson]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

+1

Deward, late's take this rare moment and celebrate. You and I have found something we completely and absoultey agree upon!

I sim, then measure flat.... From there I voice for months until I think it sounds right. Then I re-measure and post the results. They are never very pretty, but noone has yet to argue the final result.

The best example is the Fatbelly's. It seems crazy to have a huge dip at 1K which is caused by inverted phase on the tweeter and x-over points spaced slightly further apart than ideal, but they receive high praise. The strangest thing: If you normalize the tweeter's polarity, the response is almost ruler flat yet they become forward and sterile.

[fjhuerta]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

Thanks for all your thoughts.

I think of cheating as deviating from flat on axis in order to

1) pseudo-correct a poorly thought out crossover;
2) minimise the effect of choosing bad drivers.

I suppose if 1 or 2 doesn't apply, then it's not cheating at all... who knows. Maybe option 2 is not cheating, either.

[NickJ]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

I don't think it is cheating, they are for you after all, and in that sense I think nothing short of magic or drugs would be cheating in my book.

In addition to what others have said I have two thoughts. First, with the flat crossover could still get it good sounding by dropping the tweeter frequencies 1.5 with an EQ? If it still isn't right, that padding may be doing something else important.

Second, hearing is not flat or uniform. Many people tend to have selective hearing deficiencies in the frequencies of their wife's voice (best research finding ever, although I think my wife didn't like it, but I can't be sure since she is so quiet). Anyway, maybe your hearing is more sensitive than most in those ranges and you aren't actually hearing flat as flat.

[rpb]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

Stuff the mid chamber with lots of rigid fiberglass, and see if that takes away the brightness. I have a peerless driver that behaved similarly, and stuffing the chamber was what was needed. Apparently, internal reflections were getting past the cone. It had some stuffing. It needed more.

[Jeff B.]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

Thanks for all your thoughts.

I think of cheating as deviating from flat on axis in order to

1) pseudo-correct a poorly thought out crossover;
2) minimise the effect of choosing bad drivers.

I suppose if 1 or 2 doesn't apply, then it's not cheating at all... who knows. Maybe option 2 is not cheating, either.

How did you determine that this single measurement was the standard you had to go by?

[fjhuerta]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

How did you determine that this single measurement was the standard your had to go by?

That's a good question, Jeff.

I've been listening to this speaker for a year, and if someone asked me to draw how I thought it'd measure on an RTA, I'd have said there was a lack of power in the midbass and extra power around 3 KHz. The RTA measurement at my seating position corroborated my thoughts to a certain extent (the speaker had other issues, such as a diffraction peak at 2 KHz and so on).

There's one other thing. The room it is in. It's all glass and ceramic floors, plus hardwood furniture. There's very little to no absorption and a bit of diffusion (exposed bricks). So I'd assume that any trace of harshness would be amplified quite a bit.

I've checked the usual subjects: impedance, damaged drivers, incorrect measurements. So far, everything seems OK.

I'm also not very fond of any excess energy above 2 KHz.

I'm assuming the problems will be fixed by dropping the tweeter level... all things considered this solution seems to be the quickest and most effective one.

[rpb]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

That's a good question, Jeff.

I've been listening to this speaker for a year, and if someone asked me to draw how I thought it'd measure on an RTA, I'd have said there was a lack of power in the midbass and extra power around 3 KHz. The RTA measurement at my seating position corroborated my thoughts to a certain extent (the speaker had other issues, such as a diffraction peak at 2 KHz and so on).

There's one other thing. The room it is in. It's all glass and ceramic floors, plus hardwood furniture. There's very little to no absorption and a bit of diffusion (exposed bricks). So I'd assume that any trace of harshness would be amplified quite a bit.

I've checked the usual subjects: impedance, damaged drivers, incorrect measurements. So far, everything seems OK.

I'm also not very fond of any excess energy above 2 KHz.

I'm assuming the problems will be fixed by dropping the tweeter level... all things considered this solution seems to be the quickest and most effective one.

Maybe cross the tweeter higher, and leave a dip in the on axis (and off axis)response.

[Pete Schumacher]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

The voicing is the "art" behind the science of making a nice sounding speaker system. It's your job to make these things play together to make pleasing music, and you simply cannot leave that up to a microphone alone.

Don't think you need to take the art out of the process.

[JasonP]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

It strikes me that the magic voicing is what makes this hobby so great. You end up with the sound you want, not what some engineer decided you want and could be made profitable. I fell into the "lets make it flat" trap with my first design here. I find with my second, I just worried about crossover slopes and then tweaked levels by listening and EQ testing. I ignored parts of the graph that weren't flat, but attempted an overall balance in the design before voicing. I'm liking these speakers much more than the first "lets make it flat" attempt.

Just my two cents.

[fjhuerta]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

Don't think you need to take the art out of the process.

I wish I could, though. The subjective part of the hobby is the one I dislike :( I'm not very comfortable being creative, so to speak. I'm more comfortable with tried and true solutions.

[Pete Schumacher]

Re: "Voicing" or "tuning" a speaker after a simulation... cheating?

I wish I could, though. The subjective part of the hobby is the one I dislike :( I'm not very comfortable being creative, so to speak. I'm more comfortable with tried and true solutions.

But only YOU know what you "really" like.

The subjective part helps discover things like inadequate midrange chamber stuffing, simply because you're unhappy with the result, despite what the mic is telling you. That again, is the art.

Lot's of guys made swords . . . very few make a katana.