MT-32 vs. CM-32 et al.

[Tom]

I've been doing some digital recording from my Sound Canvas's and MT-32, and trying to reduce noise level that's often associated with the MT-32.  I borrowed a CM-32L from a friend to help eliminated the 'hiss' often heard with several MT-32 sounds...particularly, Soundtrack and Wind Instruments.  I've never noticed until now how UNLIKE the two L/A devices are!

The CM-32's "whistles" have a lot of built in vibrato that completely changes the character of the sound.  Even the CM-32's 'timpani' sounds more...hollow; not as rich as the MT-32's.  Overall, the CM-32 just doesn't sound as warm to me as the MT-32.  Has anybody else noticed these kinds of differences between the MT-32 and later L/A models?  I really don't feel comfortable replacing a noisy MT-32 with a CM-32L because it alters the character of the song quite a bit.  I've also noticed the Master Volume of the CM-32L has to be reduced by 50% to match the level of the MT-32 at 100% output.  This seems pretty strange, too.

[shad0wfax]

I've had both a MT-32 and a CM-32L and oddly enough, I haven't noticed much difference either in noise levels nor sound character. Maybe that's because the MT-32 was not one of the early ones. Given that, I'd keep the CM-32L. Can't say a word about early MT-32's, though.

[Cloudschatze]

The vibrato issue is discussed in several threads. If you run a search, they'll turn up.

I wasn't aware that it affected the CM-32L to such a large degree though. The vibrato of the LAPC cards and CM-64 units I've tested with has been unremarkably similar to that of the MT-32. With the CM-500 and CM-32LN, the difference becomes extremely noticeable.

Just for grins, what's the serial number of that CM-32L?

[HondaSiR]

Tom, just out of curiosity...how does the LA part of the CM-32L compare with your CM-500's LA? Which one do you prefer if there are differences?

[BlueMax]

One of the big differences is also that the CM-32L has several different reverb modes, rather than the MT-32's one.

I'm not sure what the default reverb mode is again, but I do know it produces LESS reverb than the MT-32 does.... again, by default.  that could make the difference in the sound you're hearing.

[Tom]

I no longer have a CM-500.  I'll check the back of the CM-32 when I'm done using it; right now, it's too difficult to see anything with the way it's connected.  But I have a feeling this isn't a standard CM-32 ... that it's been modified or something.  I had a CM-32 a while back for a short time, and IIRC, the sound was pretty much the same as my MT-32.  I surely don't remember it sounding like this.

[Alistair]

I'm usually pretty picky about how things sound- that's why not so long ago I did a clean out, replacing my SC-55 with a SC-55 mk II, and replacing all my old (crappy!) Sound Blaster cards for nice ones.

I remember NewRisingSun telling me a CM32L was better- but frankly, I don't want something that sounds *different*- I want my MT-32! That's what the games were composed for, and that's what I want to use to play 'em and record with.

Courtesy of Zemus, I found how to eliminate any noise or hiss from WAV  MIDI recordings, using Cool Edit 2, so it's entirely moot for me.

- Alistair

[NewRisingSUn]

But I have a feeling this isn't a standard CM-32 ... that it's been modified or something

The CM-32L shouldn't have any more vibrato than the old-type MT-32. If yours does, maybe they used the CM-32LN's control ROM with some late CM-32Ls. Maybe you can open up your CM-32L and take a look inside, noting the version number on the ROM chips.

I remember NewRisingSun telling me a CM32L was better

I'm sure those weren't my exact words. The CM-32L performs better under heavy loads of incoming data and has overall cleaner sound, because is utilizes the DAC's full 16 bits, unlike the old-type MT-32's 15 bits. The new-type MT-32 is exactly like the CM-32L minus the sound effect samples.

Most PC games were composed for the CM-32L/LAPC-I, not for the MT-32. Only Sierra, and maybe a very few others, limited themselves to the MT-32. And the different-sounding is usually minimal; the sound samples I posted a while back are pretty much the only instances I can actually find.

The CM-32LN sounds even cleaner than the CM-32L because its different DAC is factory-trimmed, removing the differential linearity error completely. Unforunately, it has the latest control ROM which causes the annoying vibrato.

The CM-500 is like the CM-32LN, except that you've got both the GS and LA parts' noise floor simultaneously.

Courtesy of Zemus, I found how to eliminate any noise or hiss from WAV MIDI recordings, using Cool Edit 2, so it's entirely moot for me.

I hope you realize that removing broadband noise always removes some part of the music signal as well.

[Cloudschatze]

The...

About time you showed up.

[Great Hierophant]

I'm sure those weren't my exact words. The CM-32L performs better under heavy loads of incoming data and has overall cleaner sound, because is utilizes the DAC's full 16 bits, unlike the old-type MT-32's 15 bits. The new-type MT-32 is exactly like the CM-32L minus the sound effect samples.

So, if you piggybacked an additional ROM onto the existing ROM of the new-type MT-32 , would you have the ultimate in LA synthesis (i.e. an MT-32 with the CM-32L sounds) or would you need to change the control ROM as well?    

The CM-32LN sounds even cleaner than the CM-32L because its different DAC has a built-in trimpot, removing the differential linearity error completely. Unforunately, it has the latest control ROM which causes the annoying vibrato.

In this case, would replacing the CM-32LN control ROM with the CM-32Ls (or a burned EPROM containing the code) eliminate the annoying vibrato?

[Tom]

Courtesy of Zemus, I found how to eliminate any noise or hiss from WAV MIDI recordings, using Cool Edit 2, so it's entirely moot for me.

I'd be curious to know what you're talking about with this.  I've used Cool Edit 2000 for years (which I believe, has several more filter options than Cool Edit 2) and have never found an acceptable manner to remove noise and hiss without affecting the fidelity of the sound...sometimes, dramatically.  I've used noise reduction filters very sparingly only if I really, really, REALLY have to, and only with limited settings.  Like NRS said, you're removing some of the music -- and for me, that's usually not acceptable.

[Zemus]

I'd be curious to know what you're talking about with this. I've used Cool Edit 2000 for years (which I believe, has several more filter options than Cool Edit 2)

Actually, it was the other way around. CE2 was the pro version of CE2k. The current version merged both and is called Adobe Audition 2.0 since Adobe bought Syntrillium a few years ago.

When the noise is loud enough, the filters do affect the sound, but with the noise levels my CM-500 produces I can usually lower it, not eliminate entirely, without affecting the sound in any noticeably manner. That's what I showed Alistair.
When it gets any louder than that, like the background noise of my mic, the filters affect the sounds a lot.

[Alistair]

I use the filters Zemus showed me in the Eco Quest 2 CD.. the sound output was identical minus hiss. I couldn't tell any difference (apart from the no hiss)- whereas I could for example with MP3 v OGG.

- Alistair

[Tom]

I have Cool Edit, Cool Edit Pro, Cool Edit Pro 2, Cool Edit 96, and Cool Edit 2000.  (My brother's given me them all.)  I didn't realize Cool Edit Pro 2 was the best version, though.  Now that I look, I see Pro 2 WAS a 2002 release.  Funny, Cool Edit was always less that $100...until Adobe took it over a few years ago.  Now...what is it...$350 or something?  Typical Adobe pricing.

So, what are the actually settings you're talking about?  I want to check this out.  I've used Cool Edit Pro 2's 'Hiss Filters' any number of times, and I typically notice that some of the 'high end' sound is removed; the sound becomes slightly flatter - not as vibrant.  (You can do the same thing with Cool Edit 2000, though it's all manual; not as many pre-defined filter settings.)

By the way, I don't think I've ever seen "Cool Edit 2".  Is that the same think as Cool Edit Pro 2?

[NewRisingSUn]

About time you showed up.

There's also that three month-old email you have never replied to... :roll:

So, if you piggybacked an additional ROM onto the existing ROM of the new-type MT-32 , would you have the ultimate in LA synthesis (i.e. an MT-32 with the CM-32L sounds) or would you need to change the control ROM as well?

Piggybacking with ROM chips? Not only would you need to change the control ROM (that instructs the LA chips to actually look for the samples), you'd also have to add at least one additional data line from the LA chip to the new ROM, and possibly modify the others as well.

In this case, would replacing the CM-32LN control ROM with the CM-32Ls (or a burned EPROM containing the code) eliminate the annoying vibrato?

I've wondered about his myself. Theoretically it should work.

[Great Hierophant]

If replacing the control ROMs of the CM-32LN would not work, would it be possible to isolate the code in the control ROM and fix it, then reburn the ROM?  It would be interesting to do a byte-comparison with the two ROMs.  The CM-32L's has been dumped and is available, so perhaps someone could provide the CM-32LN's.  Hopefully, this fix will also serve in the more common CM-500's.  I wonder if the LAPC-N also suffers from the vibrato problem, but as the control ROM is probably a soldered ROM rather than a socketed EPROM, that may be difficult to figure out.

[Zemus]

So, what are the actually settings you're talking about? I want to check this out. I've used Cool Edit Pro 2's 'Hiss Filters' any number of times, and I typically notice that some of the 'high end' sound is removed; the sound becomes slightly flatter - not as vibrant. (You can do the same thing with Cool Edit 2000, though it's all manual; not as many pre-defined filter settings.)

By the way, I don't think I've ever seen "Cool Edit 2". Is that the same think as Cool Edit Pro 2?

The hiss reduction filter seems to work more like the noise reduction system on old tape decks that removes most of the high frequency sounds to reduce the hiss. I've always used "noise reduction" when removing the background noise. You can add a profile for just the background noise and then remove a certain percent of that. It also has more options. I usually run it with FFT size set to 8192 points, transition width to 3-6dB (smooths out the cutoff point), and noise reduction level to about 93%. This won't remove the noise completely, but will lower it enough to become much less noticeable.

"Cool Edit 2" in my previous post is "Cool Edit 2 Pro". Small typo there.
I think it was the last version Syntrillium made before being bought by Adobe.

[Tom]

Thanks...I'll give this a try tonight.

[Alistair]

I think it's called "Cool Edit Pro 2", to clear up the confusion. I've been using it ever since I started editing WAV's.. 4 years ago now. (Whew!).

I now have 2.1, though, after finding an upgrade on the web somewhere.

Anyways- Yes, I know the actual hiss filters change the sound- they're awful.

Zemus' strategy means you analyse each individual WAV file and work out how much noise there is, and then CEP gets rid of it based on the WAV file. You only select about 10 seconds of "non-music" track- so when you record a 2 minute MIDI, assuming you hit record and then play the MIDI about 2 seconds after, you shouldn't stop recording until about 2:12. Does that make sense?

You then analyse the 10 seconds (or 8/9) at the end, making sure the last note doesn't resonate. I'd usually record 15 seconds silence at the end and only use the last 10. Select the last 10 with the mouse and analyse it using Zemus' settings- works for me.

- Alistair

[Cloudschatze]

In this case, would replacing the CM-32LN control ROM with the CM-32Ls (or a burned EPROM containing the code) eliminate the annoying vibrato?

I've wondered about his myself. Theoretically it should work.

There are enough hardware differences between the two that supplanting the ROM might be unsuccessful. Modifying the code might work. What would you look for though?

I seem to recall the LAPC-N ROM being soldered. It's probably the same situation for the CM-32LN as well, but I'll double-check later.

It may be possible to solicit Roland's help. Roland Japan has been extraordinarily outgoing, and I have the contact information for the lead technician at Roland US. If they can be presented with enough specifics, it might be simple enough to get a revised ROM image produced.

There's more than just the vibrato issue though, isn't there? Was the cause of the "Space Alvin" patch differences ever determined?