I recently spent a pile of cash on new home stereo gear (VMPS RM-2's and 626's), and I've decided to get serious about configuration (for the home) and equalization (for the car).

To that end, in a fit of Hack Attack, I took a cool tool called Sonic and hacked up some frequency sweep generators, including warble tones, since that seems to be something the stereophiles care about. Here's the source code, which should compile and run just about anywhere (I was doing it under cygwin on Win2K).

Questions for everybody here, now that I've got this tool:

- If you could have MP3 files with any frequencies, sweeps, warbles, and so forth, what would you want?

- What are the center frequencies for the RioCar equalizer in its default 5-band or 10-band configurations?

- What algorithm would you use, given the aforementioned test tones and a Radio Shack SPL meter, to tune your car stereo? Your home stereo?

- Are there any web sites out there that explain how to do all of this? I've looked and found nothing particularly enlightening.

Thanks!

Incidentally, it's a blast playing with Sonic. All the buggy versions of my program that I went through gave me all manner of 1950's cheesy sci-fi sound effects.

The first thing that I would suggest is making any test tones be wavs, not mp3s, in order to not have to worry about encoding issues.

Beyond that, you're over my head.

I was hoping Tony "Award Winning Car Stereo System" Fabris would join in at some point and enlighten us on this. I can now make all kinds of tones. I just need to know which ones to make.

I was hoping Tony "Award Winning Car Stereo System" Fabris would join in at some point and enlighten us on this. I can now make all kinds of tones. I just need to know which ones to make.

Hey, I don't have an award-winning car stereo. You're confusing me with Doug.

And I can create test tones in CoolEdit any time I want (and have). I've found that without a piece of spectrum analyzer software and a calibrated microphone, all the frequency sweeps in the world don't help. I get better results simply from listening to the stereo with my regular variety of music and making occasional small EQ tweaks.

I do have the itch to do some RTA passes on the system. But for that, you need the RTA software to generate the test tones (as I understand it).

Hey, here's an idea. Does anyone coming to the May 25th San Francisco meet have RTA gear? Can you bring it?

I found SpectraRTA while looking for some RTA software. They have a 30 day functional demo download. I looks like it will do the trick, along with a reasonably inexpensive calibrated microhpone. I plan on doing some serious tuning this weekend using some of the 20-20k pink noise tracks from the Autosound 2000 CD. I already baselined my PC sound card by playing the wav file through the sound card (SB Live Platinum), and it was surprisingly flat! I'll keep everyone posted on my results...

Cheers.
Phil

Hey, I don't have an award-winning car stereo. You're confusing me with Doug.

Oh, sorry. I guess I should have said award-winning car stereo FAQ writer or something along those lines.

OMFG, that SpectraLAB software is awesome.

The only things I need now are that Rat Shack microphone and to figure out where the SpectraLAB software saves its temporary noise-generation files before it plays them...

to figure out where the SpectraLAB software saves its temporary noise-generation files before it plays them...

Never mind, there is a "save signal to wave file" option. Rock.

Decisions... Do I buy the ratshack mic, or see if anyone I know owns one...

I think you should be able to find some calibration files for the radiohack mics. ISTR seeing them somewhere...

The Mighty-Mic II I ordered just arrived today - can't wait to actually try it out...

Happy tuning!

Phil

The software I just downloaded comes with a calibration file for the ratshack mike. I'm just trying to decide whether I want to shell out the 40 bucks for the thing or try to use one of the mics I've got.

Well, I've got the Radio Shack mike / analog meter. It's "good enough", and it's a lot less money than a "calibrated" microphone. It's even got a tripod socket on the bottom so you can set it up and get out of the way. I was originally thinking I'd do all of this with test tones and watching the little meter swing back and forth by hand. But if everybody's raving about this software, then I guess I should give it a try. But, I think I'll wait until I have a working laptop.

(My laptop screen is currently dead, the fourth failure of this Omnibook 500 since I got it a year ago. HP let me take 75% of the value of the old laptop toward purchase of a brand new one, so I have an Omnibook 510 (with built-in wireless Ethernet) on its way.)

Which version are you downloading? Can I use my digital VU meter (RadioShak) as a Mic? Can I find what mikes are known by this software?

I was originally going to borrow my wife's laptop for the job, but then I realized that moving my main PC out to the garage temporarily would be just fine. It doesn't need to sit in the car, only the mic needs to sit in the car. I already have the PC outfitted for taking it to LAN parties regularly (handles on the case and monitor), so it's really no big deal to lug it out to the garage.

Aside: I've refined the system so that setup/teardown of my PC is just a couple of minutes each time I go to a LAN party. I have extras of every necessary cable packed into a backpack, so I leave the cables all plugged in at home and just take the pre-packed backpack with me. The only loose items I need to bring are the keyboard and mouse. The mouse drops into the backpack and the keyboard velcros to the side of the PC case. Self-powered speakers and cables/powersupply for the empeg are pre-packed in a gym bag. Everything has a handle, so carrying things to/from the car is usually a 1-trip deal (two if I'm tired). Last LAN party was a blast. Didn't think that Medal Of Honor would be fun in Deathmatch, but we actually had a huge amount of fun with it.

If you know the response curve of your microphone (a printed graph of which is included with any mic worth its salt), you can enter the calibration data yourself by hand. The files are very simple ascii text files.

The SpectraLAB software comes with pre-made calibration files for a few mics. They are:

AKG 1000
EV 257B
Shure SM57
Shure SM58
Radio Shack Mic (part number 33-3007)

I'm set, because I have a friend who owns several SM58's. /me is going to be doing some RTA here pretty darn soon. I'm excited.

The calibrated file for the radio shack mic looks like this:

; Radio Shack low cost condenser mic (P/N 33-3007)

0	-50.0

40	-7.5

50	-7.0

80	-6.0

100	-6.0

120	-5.0

200	-3.5

300	-2.8

400	-2.0

1000	0.0

1100	0.8

1300	0.2

2000	1.0

3000	1.5

4000	1.8

5000	2.0

6000	2.2

7000	2.5

8000	2.4

9000	1.0

9500	0.0

10000	0.0

12000	-0.5

13000	-4.0

15000	-3.0

18000	-2.5

20000	-3.0

I just wanted to say how so far beyond me this is.

Anybody feel like dumbing down the conversation for somebody who doesn't understand all the technical jargon.

Although I have my Pioneer as my main HU, it has an advanced EQ that includes adjustments for the Q. I'd love to set it up better if I had the technology and knowledge to do it. Right now I use their defaults.

Also, using a sound meter, you could create a constant tone that only uses the Left or Right speaker that you could then use to adjust left/right in the car. But, isn't it impossible to separate Front/Back?

Anybody know how much it costs? Thirty days is nice, but I could get some use out of this (though I'm what's professionally known as a 'cheap bastard'.)

Greg

Greg,

I had the same thought. I e-mailed to get the price list and it looks like:

RTA132A Base Analyzer... $295.00
132B ValuePAK: base + 3 options .....$ 795 (You Save_33%)
132B+ ValuePAK: base + 6 options ...$ 995 (You Save_52%)
132C ValuePAK: base + 10 options ...$1495 (You Save_54%)
Add-On Options ... $295/each
Optional Scalable Plug-&-Play Hardkey Dongle (parallel port): $100USD

I don't know what all those options are, but I can't justify spending the $295 for the base.

FWIW, I used my $50 Radioshack digital SPL meter with those Autosound tone tracks (converted to WAV) to do some preliminary balancing and it helped. It was pretty crude, but it helped get rid of some glaring peaks and lows (in bass ranges mostly) that I fixed with subwoofer gain and filters before I started in with the equalizer and and incremental fine-tuning by ear.

If I stumble across $295, it would be even nicer to use this product, but I'm happy enough for now.

Damn, I'd better get my RTA work done pretty quick before my demo copy expires, then!

It's OK, I've got the SM58 right here in my hand ready to go this weekend.

Anybody feel like dumbing down the conversation for somebody who doesn't understand all the technical jargon.

I'll have a go at this and see what I can do. My beef is that I've been unable to find a decent FAQ out there to explain all this garbage.

The core goal here is to have a tool that tells you the frequency response of your system, which is to say, which shows you where your system isn't flat. The graphs produced by SpectraRTA and other such programs have frequency on the X axis and loudness on the Y axis, and then they make frequency sweeps from your sound card, measuring them with the microphone.

Of course, just like speakers, microphones aren't flat either. Thus, good mikes come with a "calibration curve" telling you which frequencies they tend to get low or high. SpectraRTA tries to correct for this.

Then, you're supposed to tweak your equalizer to fix this. Tony has a nice entry in the RioCar FAQ about Q parameters and all that. Basically, you look at the the graph given by SpectraRTA, you make a bunch of tweaks to your equalization settings, then you repeat. Ad nauseum.

Other terms I've seen pop up here:

Warble tones: if you use a "pure" sine wave, this might set up resonations in your car which are louder than what the speaker is actually putting out, causing you to over-compensate when you tweak the equalizer. Warble tones take the frequency of a pure sine wave and move it up and down, maybe five times a second. This, in theory, helps you get the "true" frequency response of your system.

Pink noise / white noise: There's a nice web page that explains these terms. Roughly speaking, pink noise has the same amount of "energy" at every frequency in the audible spectrum.
You can also use pink noise to test for system flatness, although frequency sweeps seem to be better for measurement.

Okay, here's a pretty basic question, then (which although it is interesting, the article you just linked does not answer this question)...

If I played a noise waveform, and used SpectraLAB to adjust my EQ to "perfectly flat", which noise waveform should I use? Pink noise or white noise?

Would the options be additional licenses? Hrm, I wonder if an educational discount makes it free? Too bad it's dongled... A price like that would drive an occassional user to piracy. ... I've said too much.

Greg

I believe pink noise is the way to go. Since it's contant energy, a properly adjusted eq will result in a flat curve. White noise will result in "wedge" shaped curve, ie climbing higher as frequency goes higher. I played around with the built-in pink noise generator, but I think I'm going to stick with the Autosound 2000 pink noise track as it appeared to be a bit flatter when played as a wav on my PC. My goal this weekend is to use their SpectraRTA software (the other versions appear to be overly complex and not as responsive) and the Autosound track ripped to a wav.

Cheers!