Mac revisited

[asherbro]

Hi everyone,

I've been working with REW under Mac OS X on a PPC machine. I've seen some comments here about anomalous results with the Mac audio subsystem, due to their incomplete Java support, but others indicating that people were getting the program to work there, so I thought I'd at least post.

I have J2SE 5, and I've been working with versions 4 and 3.28 of REQ. I can set things up so I have I/O working, but with both versions I get "Impulse Peak is not where it should be, the measurement may have been corrupted" when I run a measurement and, more bizarrely, I get a really strange tone from version 4.0, during measurement. It's not the orderly sweep I would expect based on my experience with 3.28 (where I was also encountering the clicking during all pink noise tests, hence the upgrade to 4, which has at least corrected that). It's this weird sci-fi laser sounding warbly sweep <g>. Has anyone seen this? Is 4.0 working on anyone's Mac OS X machines? Thanks so much!

Sincerely,

Andrew

[Josuah]

I got some "measurement bad" messages when I was using the built-in mic/speaker inputs and outputs, both on Windows and Mac OS X. When I switched to an outboard USB pre-amp (by M-Audio) those errors went away.

[JohnM]

I've started to look at REW on the Mac, it's kind of depressing how many basic rendering and display formatting issues there are on top of the audio issues given the basic premise of Java as being largely platform-independent, but I'll press on with adapting REW to the Mac in any case and advise when there is a build that works better under OS X.

[asherbro]

>I got some "measurement bad" messages when I was using the built-in mic/speaker inputs and outputs, both on Windows and Mac OS X. When I switched to an outboard USB pre-amp (by M-Audio) those errors went away.

Thanks - I've tried my Echo Mona PCI card, as well as the built-in audio system, with the same results.

>but I'll press on with adapting REW to the Mac in any case and advise when there is a build that works better under OS X.

Thanks so much, John!

asherbro wrote:

Hi everyone,

I've been working with REW under Mac OS X on a PPC machine. I've seen some comments here about anomalous results with the Mac audio subsystem, due to their incomplete Java support, but others indicating that people were getting the program to work there, so I thought I'd at least post.

I have J2SE 5, and I've been working with versions 4 and 3.28 of REQ. I can set things up so I have I/O working, but with both versions I get "Impulse Peak is not where it should be, the measurement may have been corrupted" when I run a measurement and, more bizarrely, I get a really strange tone from version 4.0, during measurement. It's not the orderly sweep I would expect based on my experience with 3.28 (where I was also encountering the clicking during all pink noise tests, hence the upgrade to 4, which has at least corrected that). It's this weird sci-fi laser sounding warbly sweep <g>. Has anyone seen this? Is 4.0 working on anyone's Mac OS X machines? Thanks so much!

Sincerely,

Andrew

Just tried 4.0 for the first time today. I got the exact same results as Andrew.

Mike

[paul99]

Hi John -

I'm new also, just trying REW 4 on the Mac (a G4 power book). I get the same problems as above.

What I hear is that somehow the output test signals are way overdriven into clipping. When I use the generator panel, I have to lower the output level to -60 dB to get an audibly clean sine wave or sweep. But in the real measurement section, the lowest level I can set is -40 dB, which distorts badly.

The badly distorted output is what gives the "measurement bad", or "impulse peak not where it should be", cause there's no way to get a decent sample from this signal.

I have tried using the Line inputs & outputs on the laptop, as well as a decent firewire audio interface (the Metric Halo Mobile IO 2882). Both have exactly the same issue, so this is probably a Java->CoreAudio problem, as you suspect above.

Is there any testing that I can do, to help you track this down? Just name it and I will try....

The program looks awesome, and I'd sure love to have it working on the Mac. I guess in the meantime, I will try to set it up on a PC.

Thanks for all your efforts!!

paul

[JohnM]

Thanks Paul, that gives me some useful pointers.

Hi,

Another Mac user here! I have the same problem with REW 4 as described in the first post (great description, BTW). I'm running a G4 PowerBook with system 10.4.8 with all the latest updates, "Java J2SE Release 5".

REW 3.29 runs just fine (as far as I can tell). The measurement sweep sounds just fine. While I find the interface of REW 4 much better I'll have to stick with 3.29.

I, too, would be willing to try to help troubleshoot REW 4 for Mac.

Mace

Ok, I'll just piggyback on this. I'm getting the exact same result with version 4, newest Java and OS 10.4.8. I'd be willing to beta test if needed.

Michael

[thxgoon]

Does anyone know if there's a way for REW to run on 10.3.9? Old versions?

I'd like to know this, too. Anybody?

C

[BoomieMCT]

I have used my fiancee's G4 notebook for readings (easier then moving my desktop PC around). I too have found the Mac experience somewhat . . . wanting. My list of troubles include;

1) The Mac soundcard only does the lowest sampling rate.
2) The Mac soundcard introduces some odd high freqency anomalies.
3) The Mac soundcard has trouble with the volume controls. I also have the clipping problems paul99 noticed with many setups.
4) The Java interface has trouble with the graphics and produces a lot of graphics artifacts. This is actually the most annoying "feature".
5) REQ on the Mac locks up a lot. Actually, it has been my experience most programs lock up a G4.

I would appreciate anyone's solutions to these problems. I've been using version 3.29 - I'm kind of scared of 4.0 on a Mac.

As to whether you should upgrade to REW 4.0 with a Mac - don't. It just doesn't work at all, you can't run a sweep.

I have also seen the strange high frequency anomalies with REW when i do the soundcard calibration. Not sure if it is the Mac Java or REW or the Mac soundcard.

I haven't experienced the distortion problem with my set up with REW 3.29 on my PowerBook G4.

As to your G4 locking up that is strange. My PowerBook G4 is up for months at a time with the only restarts being after system updates. It is virtually crash free. If your Mac is locking up with most programs there must be some system issue. You could try repairing disk permissions which may help overall stability.

Mace

Can anyone tell me if I can use any version of REW if I'm running OS 10.3.9? Thanks,

C

I'm about to buy a SPL meter to try this on my G5 iMac running 10.4.9.

There's also a program called FuzzMeasure which seemed pretty cool to this layman, except I doubt I was getting accurate results using the built-in mic.

I'm unclear on why FuzzMeasure recommends using an expensive high-quality omnidirectional mic while REW users seem to get by with the Radio Shack SPL meter. Is it because REW users are mostly concerned with the lower frequencies?

--mj

[brucek]

I'm unclear on why FuzzMeasure recommends using an expensive high-quality omnidirectional mic while REW users seem to get by with the Radio Shack SPL meter. Is it because REW users are mostly concerned with the lower frequencies?

That's certainly partly true, but we also recommend REW users get a Behringer ECM8000 omni-directional microphone with its associated preamp. Another choice is the Galaxy CM-140 meter (we also have a cal file for the Galaxy).

Both those choices are quite good and are considered far better than the RS meter. We supply a cal file for the RS meters because the reality is that most people don't want to afford the better microphone or meter, so the RS meter is better than nothing...

brucek

Just an update that the tone generation problem in 4.0 affects my G5 iMac also. The 1KHz test signal for calibrating the sound card seems extremely distorted, as if run through a ring modulator like Dr. Who's Dalaks. The tonality changes when adjusting the volume, suggesting clipping somewhere. I tried this with the system audio, an Eridol UA-20 outboard audio interface, and a USB headset.

Another Java 5 app I use on OS X works well, though it isn't doing any audio.

I'll try with 3.29.

--mj

3.29 working fine so far.

--mj

Update: I find the new UI so much better than the old one I've tried running the new version on Parallels (virtual MS-Windows) on my Intel-based MacBook plugged into the Eridol UA-20 USB audio interface.

REW's menus work properly, but the Eridol doesn't seem to work with MS-Windows, despite having the latest XP driver. I don't know whether this is related to running MS-Windows through Parallels.

I guess I'll fall back to 3.29. I'm also curious about FuzzMeasure, but I was really impressed that REW 3.29 detected clipping during my initial sweeps, and not sure FuzzMeasure has features like that.

--mj

The good news is the REW tone generation works fine on my Intel MacBook. Maybe the problem was limited to PPC-based Macs like my G5 iMac.

I've attached the loopback test screen. Does this frequency response look good enough for audio testing?

The bad news is that so far for me it only works with the built-in audio hardware, not my outboard Edirol UA-20. This will work with the Radio Shack SPL meter (using an RCA to 1/8" adaptor), but not my measurement mic that needs phantom power. I don't know whether this is a problem with REW, OS X, Java, or the Edirol device.

--mj

[JohnM]

That response looks fine. In principle you should be able to leave REW set to Default Device for input and outpout and change the device to use for input and output in the Mac Sound panel, but on my Intel Mac Mini I've found that tends to crash the Java Runtime Environment.

I've fixed most of the display/rendering oddities on the Mac and sent the Menu bar to its traditional Mac home, those fixes will be in the next release, but it looks like getting the audio interfaces working properly with devices other than the built-in hardware will mean changing to Apple's CoreAudio class library, which isn't a 5 minute exercise as the API differs from that used by JavaSound.

FuzzMeasure looks to be a very well developed Mac application and the trial version is free, I believe, so well worth investigating.

More bad news. REW 4 works correctly on my Intel MacBook, then crashes when taking subsequent measurements. I suspect it's something simple like running out of memory. I'm about to try launching from the command line with -Xmx500m or so.

Changing the default input from the System Preferences wasn't seeming to affect which source REW used. It kept using the built-in mic instead of my Edirol thing (which may well be a bug in the Edirol driver for all I know).

Here's one of the sweeps I captured anyway. Looks like a big hole at 96Hz.

[brucek]

Looks like a big hole at 96Hz

Fairly narrow. You'll likely not ever notice.......

[woody77]

Another mac user here. I have both an Intel MacBook Pro and a PPC PowerBook. It's pretty much unusable on the PPC box, my guess from listening to the sounds was endianness of the data.

If you're using a java nio byte-buffer for the data, that might be an issue. But I'm glad I have at least one box that works.

[mrsollars]

going to be trying to use an intel macbook. any suggestions??

thanks