Difference in melody playback between midi and realtrack backing

PG Music Home

I'm having a problem with changes in melody playback when using an all midi vs. all realtrack backing track. Here's a link to hear the problem: http://www.soundclick.com/bands/default.cfm?bandID=1111184. The first 12 bars used the "Cool_One - Slow Sw. Jazz guitar" midi backing, while the second 12 use _J085_GB "Cool Jazz Ballad" backing track. The melody is a simple triplet, triplet, 8th, 8th, rest.

Why am I having the problem and, more important, how do I get rid of it?

Equipment: Core I5 with 16gb RAM,M-audio fast track ultra with Ketron SD2, if that matters.

Thanks for the help,

Bob aka jazzwombat

PS: "midi v rt problem" mp3 on soundclick. Listen for the "hesitation" or lack of uniformity of the triplets when the RT backing (second set) is used.

Bob,

Is the melody a midi melody or a live playing? Also, do you have the song set-up so that when the style changes it changes automatically for the second 12 bars?

If you have a style change from Midi to Realtracks, BIAB needs to generate the Realtracks and doing this requires a hefty amount of the computer's processor if "Fast generation of Realtracks" is enabled. The reason for this is that BIAB will commence playing before the Realtracks are fully generated and the completion of generation occurs as a background process. Now, if you throw into that mix the fact that you are also asking your computer to record audio, there's a lot being expected and the computer will share the processor between events. By comparison, generating a Midi-only accompaniment is fast and efficient and does not use too much computer power.

As an experiment, try deactivating the fast generation of Realtracks under "Opt | Preferences | Realtracks" and see if that makes a difference. With this turned off, the song will not commence until all Realtracks are complete so it will take longer to start up. The upside is, though, that if you are recording, pretty much all processing power can be devoted to that.

Hope these thoughts help,
Noel

Hi Noel:

It's directly from BIAB-generated notes on the melody track.

Bob

Hi Bob,

Please see my above post. I've edited it with extra information. (Just to keep everything in one place.)

Seeing you are using an external synth, it seems to me that you are recording audio at the same time that Realtracks are being generated.

As mentioned above, try deactivating the fast generation of Realtracks. Another thing you could try is to use the internal synth and see if the problem persists with that.

Regards,
Noel

How did you get the melody?

Thanks Noel. Certainly worth a try, although the strange hesitation when RT backing is used is evident even if I am not recording, as you noted. I'll also try the "return to factory default", after saving my current preferences. I'm at my office right now, so I'll make the changes and report back later tonite.

Quote:

How did you get the melody?

I keyed in the notes in the notation window for the first bar, and then copied to the next 11 with a 5 semitone increase (to go around the circle of 4ths). The notes look fine on the notation, leadsheet and in the piano roll windows.

Bob

I like the first version better. Oh, that's not what you are asking?

Yes, it does sound like the melodist has been drinking a bit the second time around.

Can you send us the actual BIAB files so we can see if the issue in your system?

Quote:

I like the first version better. Oh, that's not what you are asking?

Yes, it does sound like the melodist has been drinking a bit the second time around. Can you send us the actual BIAB files so we can see if the issue in your system?

I'd be happy to do so, but how do I get the files to this forum? There doesn't seem to be an attachment option.

Bob

Try freezing the melody track.

Here is Bob's (aka jazzwombat) file causing the weird effects .. when he uses a RT backing track, but not when using a midi backing.

I'll take a look at it a little later, in the meantime, perhaps someone can see this they can reproduce the problem.

Jeff's lick.MGU

Thanks jazzmandan! I tried freezing melody (and other instruments) without change. I restored to factory settings and that didn't seem to make any difference. I changed the realtracks options to favor a slower machine (I've got a relatively fast one) without change. I doublechecked my midi settings, and they seem to be right for my fasttrack and m-audio asio drivers. I'm left thinking it must be something with the file itself. I've played several songs that are demos of the various styles, changing from realtracks to midi styles, and couldn't replicate the problem. So, if someone has a minute to check out the file at jazzmandan's site, please let me know if you see something I've done wrong.

Thanks,

Bob

Bob,

The file plays fine for me using the _J085_GB.STY (Cool Jazz Ballad (85 RS))style. I'm using a VSt/DXi internal synth.

Are you using multiple choruses with this file and having the style change each time through?

Regards,
Noel

Quote:

Bob,

The file plays fine for me using the _J085_GB.STY (Cool Jazz Ballad (85 RS))style. I'm using a VSt/DXi internal synth.

Are you using multiple choruses with this file and having the style change each time through?

Regards,
Noel

Hi Noel,

No. What you heard on my soundclick file was two separate recordings that I merged in Audacity. The first was using the midi style for one chorus of 12 bars, and the second was the RT (_J085_GB; Cool Jazz Ballad) style for one chorus of 12 bars.

So, does this suggest I have something set incorrectly in the midi preferences in BIAB or something needs to be change to the M-audio driver settings, or what? I checked the midi settings in BIAB and they look fine, but how would I know? Any suggestions for where to begin are appreciated. Thanks.

Bob

Quote:

Bob,

The file plays fine for me using the _J085_GB.STY (Cool Jazz Ballad (85 RS))style. I'm using a VSt/DXi internal synth.

Same findings here. The midi melody playes right on beat as expected with this RT Style. Not at all like I heard in the recoding you provided earlier in this thread. BTW, I am using heh Coyote Forte Dxi synth.

Bob,

Route Melody through the Coyote WT VST/DXi and see if the problem persists. This will allow us to identify where the problem might be occurring.

Regards,
Noel

Hi Noel,

I routed thru Coyote Forte DXi and used the M-Audio ASIO driver that I used previously. If anything, I believe the problem got worse. I posted the mp3 on my soundclick page: http://www.soundclick.com/bands/default.cfm?bandID=1111184.

What strikes me as odd is the randomness of the "triplet mistake". I swear I've listened to the RT of these 12 bars 4 or 5 dozen times, and the error in the melody never truly repeated itself. I've listened to a couple of other melodies, particularly ones I didn't enter myself ("Lady Bird" was one). Believe it or not, I can't hear problems with them, although they don't tend to have the bar-to-bar repetition that is in my file, so this may not be a fair comparison.

There are a couple of things I could try. For example, I could change to the ASIO4ALL driver -- bypassing both the FastTrack and Ketron -- and see what happens. Alternatively, I could reset to "ALL" factory default settings; previously I reset to "MOST" default setting, and that didn't seem to correct the problem. What else might I try? I'm at my wit's end.

Thanks for the help everyone. I appreciate your efforts.

Bob aka jazzwombat

Interesting update:

I made separate files of the RT and midi 12-bar chart, and then listened to them in RealBand. First, both played flawlessly. Second, in each case the melody track was approximately 18 bars long and contained notes stretching beyond the 12 bar ending. I have no idea where these came from, but they are there in RB. Deleting them in RB made no difference: in each case, the triplets were replicated correctly.

Someone, please explain what the heck is going on here? I will copy and paste the 12 bars to a new file and see if that cures the problem, whatever it is.

Bob

Copy and pasting did not cure the problem. So, as it stands, both midi and RealTrack files play correctly in RealBand, but triplets are not rendered properly when RealTracks backing is used in BIAB. ????

Hi jazzwombat,

If this is possible with the Fast Track, you should try switching from ASIO to MME audio drivers in Band-in-a-Box. Just see if that changes anything.

If that is not possible, find out what the ASIO driver's buffer size is (should be able to see this in the ASIO control panel).

Hi Blake: No can do with MME. ASIO buffer size is 4096. The same ASIO driver is used in both BIAB and RealBand, so I don't feel that it is to blame. Seems something inherent in BIAB (and not in RealBand) is behind this particular problem, no?

Quote:

Interesting update:

I made separate files of the RT and midi 12-bar chart, and then listened to them in RealBand. First, both played flawlessly. Second, in each case the melody track was approximately 18 bars long and contained notes stretching beyond the 12 bar ending. I have no idea where these came from, but they are there in RB. Deleting them in RB made no difference: in each case, the triplets were replicated correctly.

I see exactly what you mean. Here is Jeff's Lick in RB:

Unfortunately, no explaination.

Bob,

I have no more ideas. Sorry. The file works fine for me so that seems to indicate that it might be something in your set-up.

Some questions spring to mind ...

1. Have you installed the latest update for your version of BIAB?

2. Have you restarted your computer from its fully off status and not suspend?

3. Does the problem persist with other RT styles?

In one of your above posts, you mentioned some other thoughts. They sounded good.

I agree with you. Try installing ASIO4ALL and seeing if it makes a difference. ASIO is a strange animal and it's caused many people issues here in the forums.

You mentioned above that you were creating a new song file. To be honest, if I was having this problem, that's exactly what I'd be doing now. I wouldn't be copying and pasting anything though as the very nature of copy/paste means it's possible to transfer the issue.

Kind regards,
Noel

Hi Noel et al:

First and foremost to all who have offered ideas and suggestions: THANK YOU! I greatly appreciate your efforts. Noel, to address your questions: Yes to all three. I think I have to look at my preferences in BIAB. Since both midi and RT files play correctly in RealBand, which uses the same hardware and driver to render audio, I conclude that BIAB is where the problem lies. So, just as soon as I can, I plan to reset everything to factory default settings, using the ALL setting in the reset option. Once reset, I'll use the ASIO4ALL driver and ForteDXi with the prediction this configuration will work. If so, I'll add back in succession the M-Audio ASIO drivers, the Ketron SD2 patch map, etc., and start changing to suit my preferences. Hopefully, through this process, I'll find what is creating the RT/melody issue. I'm somewhat comforted by knowing that the tracks work in RealBand. That at least gives me some assurance that it isn't a general hardware issue. (Ditto for Sonar X1c. No problems whatsoever with it either.)

If anyone has additional thoughts or opinions, please share them. I'll post back with the results. Wish me luck,

Bob aka jazzwombat

Hi jazzwombat,

If possible, try setting the buffer size lower - 512 would be a good place to start. You will need to close and re-open Band-in-a-Box after changing this setting.

1024, 2048, or 4096 may be a more stable setting for you, but this would just be a troubleshooting step.

Thanks Blake. I'll give it a try. Would you mind explaining why a small ASIO buffer might rid this problem? I ask simply to learn more about what's going on 'under the hood' with ASIO drivers?

Cheers,

Bob

Having an overly large buffer size (2048 being considered large) can lead to poor midi timing from the DXI VSTi synth.

Thanks, but I thought, since I use a hardware synth (Ketron SD2) and not a software synth, that wouldn't be an issue. Wrong?

Your right hardware would unaffected by ASIO buffer size. I'll see what else I can find out for you.

I changed from using the M-Audio Fasttrack ultra ASIO driver and Ketron SD2 patch map to the ASIO4ALL driver (12 ms latency), the default patch map, and Forte DXi synth. This combination _works flawlessly_ with both the midi and RT styles used previously.

I then switched to the M-Audio ASIO driver: the RT-melody problem returned. I confirmed this was not due to the Ketron SD2 patch file nor the Ketron SD2 itself. So, it appears to be a direct result of the M-Audio driver (I guess... second opinions welcomed).

The question becomes: what can I do? I have the latest driver version (5.10.0.5100; firmware 1.51). I don't want to buy another hardware audio interface or synth (the Fasttrack is less than 1/2 year old and I'm happy with the Ketron SD2); however, I am unwilling to live with this problem. There are not many things I can (as a user) to change the settings on this driver: I have options for sample rate (currently 44.1 kHz and not changeable), ASIO/WDM buffer size (currently 4096 samples; I tried going down to 512, this didn't help), clock (set to internal not changeable); "enable asio direct monitoring" (currently off; tried turning it on -- didn't help), and "effects" (currently "Echo" with duration, feedback, and volume set to minus infinity. There is "no effect" setting; the default is "Echo"). The driver panel allows monitoring of the eight channels on the Fasttrack. I have the volume sliders for channels 1-2 (mic in, when I use a mic) and channels 5-6 (Ketron in) set to max. There is a set of knobs below each channel labeled "send": channels 1,2,5, and 6 are all set to minus infinity. The master out volume is at max and its return set to minus infinity. That's all I can alter. Anyone see anything that might be wrong with these settings?

Anyone have suggestions on how to proceed?

Thanks for the help!

Bob aka jazzwombat

I'd be checking the M-Audio support page on their website to see if they haven't released an updated driver for the device.

There may also be *more than one* driver published, sometimes they leave up an older one that is known to work better with some systems as well as the latest one. So check what version number the driver you already have against any new ones posted there.

If there is a difference, try downloading and installing the new driver.

Sometimes simply reinstalling the same version driver can correct problems as well, in the event that a driver file becomes corrupted.

--Mac

But the driver works in Real Band, right ?

Mitch: Yes, you have it correct. The driver works fine in Real Band (and with Sonar X1c, as far as I am aware). Go figure. lol.

Thanks Mac. I'll be doing that on Saturday, when I have a bit more time to go over things. I know you're a big one for pressing "return to factory settings" when problems in BIAB arise. I have not done a complete reset. What do you think, would this be worth a try? Maybe before reloading the driver or trying older versions?

Bob aka jazzwombat

addendum: Just checked the M-Audio site, there is but one driver for the Fast Track Ultra, with no beta drivers listed. Well, that eliminates one possibility, although I may take the time to reinstall it, just to be doubly safe.

Bob,

It's safe to do a full reset as BIAB saves the pre-reset set-up these days. This means you can always return to what you had if the full reset does nothing.

Regards,
Noel

I think I _may_ have solved the problem. (At the very least, the real tracks backing with the original melody currently sounds fine.) Here's what I did:

I read the FastTrack Ultra Manual several times. Not the best user manual, I must say. From that, I reset the driver configuration to:

bypass on channels 1/2 and 3/4 (this removes all send to the internal FastTrack DSP)
ASIO buffer size set to 256
enable asio direct monitoring

all other parameters were left as originally indicated in my previous posts

I am at a loss to explain why these changes now have corrected the problem. So, I only tentatively offer this as a solution. I'll shutdown and restart my computer, and I'll post back what happens. Wish me luck.

BTW, Noel: thanks. Yes, I agree with you, resetting to factory defaults in 2011 is much better than in years gone by, much better.

Bob aka jazzwombat

Bob did you happen to change the actual usb plug you were using during this?

Lots of problems seem to arise when you say install slip printer 1 into usb 2,then one day flop that and the other printer around during a reconfig and nothing works until they go back to their 'home' spot. Weird or what?

Hi John,

No, I did not change the usb plug's location (physical port on the computer), but I agree that doing so frequently causes problems and the reinstallation of usb drivers.

I listened to the RT style with the melody this morning, and I'm happy to say that the fix has stuck and all sounds good. So, I have corrected the strange melody problem in BIAB. Honestly, I don't know which of the changes made the difference, and I guess I shouldn't care too much. It's working, so all is good.... for now. lol.

Bob aka jazzwombat

P.S. I will delete the soundclick files pertaining to this problem Jan 8, 2012.