now is this issue re:bpm just related to the KP3 detecting the bpm from incoming audio or is it also an issue with midi clock as well?
Bill
kp3 auto bpm woes
Moderators: Sharp, X-Trade, Pepperpotty, karmathanever
-
decrepitude
- Full Member
- Posts: 169
- Joined: Tue Dec 26, 2006 4:20 pm
- Contact:
Yes, it's really great that Jerry from Korg is monitoring these posts. Apparently there are some issues with Auto BPM.
I also conducted a test yesterday. There do appear to be some issues.
I ran a line level signal from my computer interface into the KP3 for this test. I chose several percussion loops at varying BPM's.
Usually, the KP3 was able to detect BPM's within about .2 to 1.0 bpms of the original material's tempo. For most of the beat synced FX this was just fine. Things start to get outta wack when using the sampler.
As someone else has said, it appears the BPM algo tries to detect in too short a time. Within an 8 bar loop of breakbeat style material, the Auto BPM would fluctuate within the loop, but if stopped at the end of the loop very often the BPM had settled on the actual tempo.
Turning Auto BPM off and on again seems to be very problematic. I could have pretty good tempo lock, turn off the Auto Bpm and then on again and a 132 bpm loop would be recognized as 90.
One good thing I noticed is you can help Auto BPM along if you tap out the tempo. After a few taps, it seems to get the Auto BPM algo to sit around the mean tempo again.
One technique I had success with was to get in the neighborhood of correct tempo using a combination of Tap and Auto BPM and turn off Auto BPM when I was pretty sure I had the sweet spot. I could then stay in sync for many multiple bars of material, only using the 'Align' function periodically.
Despite all these workarounds, I think it will be very difficult for DJ's using turntables to layer percussive beats loaded upon the KP3 to play in sync along with beats off the turntables. Any other material, even slightly percussive should be just fine. Again, this is of course an issue with Auto BPM, I'm not so sure using Midi Clock Sync would have these problems.
Not using sync at all, I conducted another test. I matched up BPM on the KP3 with an 8 bar prerecorded drumloop at a given tempo. I got the loop playing and sampled the drum loop and at the end of the 8-bar sample it automatically started to play back in time with the original loop. However, I could hear the phase offset on each successive iteration of the sampled loop. Fortunately, it took about 2-3 minutes of looping before I even detected any flamming. "Good enough" sync for about anything other than layering additional preloaded drum loops. (Again, this might all be resolved with the use of Midi Clock Sync and even an additional note-on msg on every downbeat on bar #1 - which usually would not be an option for a DJ using turntables.)
I also conducted a test yesterday. There do appear to be some issues.
I ran a line level signal from my computer interface into the KP3 for this test. I chose several percussion loops at varying BPM's.
Usually, the KP3 was able to detect BPM's within about .2 to 1.0 bpms of the original material's tempo. For most of the beat synced FX this was just fine. Things start to get outta wack when using the sampler.
As someone else has said, it appears the BPM algo tries to detect in too short a time. Within an 8 bar loop of breakbeat style material, the Auto BPM would fluctuate within the loop, but if stopped at the end of the loop very often the BPM had settled on the actual tempo.
Turning Auto BPM off and on again seems to be very problematic. I could have pretty good tempo lock, turn off the Auto Bpm and then on again and a 132 bpm loop would be recognized as 90.
One good thing I noticed is you can help Auto BPM along if you tap out the tempo. After a few taps, it seems to get the Auto BPM algo to sit around the mean tempo again.
One technique I had success with was to get in the neighborhood of correct tempo using a combination of Tap and Auto BPM and turn off Auto BPM when I was pretty sure I had the sweet spot. I could then stay in sync for many multiple bars of material, only using the 'Align' function periodically.
Despite all these workarounds, I think it will be very difficult for DJ's using turntables to layer percussive beats loaded upon the KP3 to play in sync along with beats off the turntables. Any other material, even slightly percussive should be just fine. Again, this is of course an issue with Auto BPM, I'm not so sure using Midi Clock Sync would have these problems.
Not using sync at all, I conducted another test. I matched up BPM on the KP3 with an 8 bar prerecorded drumloop at a given tempo. I got the loop playing and sampled the drum loop and at the end of the 8-bar sample it automatically started to play back in time with the original loop. However, I could hear the phase offset on each successive iteration of the sampled loop. Fortunately, it took about 2-3 minutes of looping before I even detected any flamming. "Good enough" sync for about anything other than layering additional preloaded drum loops. (Again, this might all be resolved with the use of Midi Clock Sync and even an additional note-on msg on every downbeat on bar #1 - which usually would not be an option for a DJ using turntables.)
jerrythek wrote:Hello folks:
Thanks for all your input regarding the Auto-BPM sensing. Korg Inc. has been looking at this based on our reporting your threads and we "think" we found a small bug in the smoothing of the BPM sensing mechanism. We're looking into it and will get back to you as soon as we have an answer. That's all I really know right now, but thanks to your input we're "on it".
Regards,
Jerry
Excellent news.
While you are at it, could you also look at the BPM smoothing that is applied to incoming midi sync, as Im finding that slaved it to an external midi clock seems to introduce alot of artifacts into the delays (ie odd squeaks as if the delay is sometimes having its time randomly modulated). The end result is its not really useable when clock slaved.
-
decrepitude
- Full Member
- Posts: 169
- Joined: Tue Dec 26, 2006 4:20 pm
- Contact:
Well then it won't come as any surprise to know that after testing it last night I too found some "gotchas" regarding midi sync. I would say first though that all these issues relate mostly to the sampler. Before we draw too many conclusions, let's remember we're still learning all the ins and outs of the device.
OK. So my test revealed a few things:
- It's no wonder that the 'Align' function is there. Even with midi sync, recorded samples eventually drift. In my estimation it takes about 64 bars before it's even slightly noticeable (the one exception being percussive loops). For those folks incorporating lots of changes, no worries. But let's say we want to bring in a recorded phrase after many minutes have elapsed (something I love to do!). No doubt you'll need to perform an 'Align' just prior to unmuting the sample.
- When the KP3 is using external sync you're unable to check the BPM setting on the KP3. Instead it reads 'EXT'. Occasionally, I'd try to take the KP3 out of external sync and see what the actual BPM was. It was consistently wrong. Even if I kept the master tempo locked, each time I took the KP3 in and out of external sync it had guessed an entirely different BPM, sometimes by several beats. Now mind you, I can't be sure this is actual evidence of a meandering, because to my ears all beat synced effects AND properly executed phrase sampling stayed in time as I described earlier. But as the previous poster indicated, there do seem to be quantifiable inaccuracies.
Auto BPM also has issues - you'll often be needing to tap tempos to help it along. And keeping recorded samples in sync over a period of time will require occasional 'align' corrections.
In my view, the KP3 sampler is basically a glorified "tape looper" unless you record one-shots. After reading (or should I say 'deciphering') the KP3 manual a bit more, it occurred to me that one workaround might be to record a loop as a one-shot (make your recorded loop length 16 bars but manually stop it at the end of 8 bars). and trigger it with a note-on message played by the master sequencer. This would of course require a dedicated track (and some pre-planning) and would essentially play a note-on message on beat one for a given pattern. It could potentially be a solution for those who want to trigger prerecorded loops that they load into the KP3 via the SD card.
OK. So my test revealed a few things:
- It's no wonder that the 'Align' function is there. Even with midi sync, recorded samples eventually drift. In my estimation it takes about 64 bars before it's even slightly noticeable (the one exception being percussive loops). For those folks incorporating lots of changes, no worries. But let's say we want to bring in a recorded phrase after many minutes have elapsed (something I love to do!). No doubt you'll need to perform an 'Align' just prior to unmuting the sample.
- When the KP3 is using external sync you're unable to check the BPM setting on the KP3. Instead it reads 'EXT'. Occasionally, I'd try to take the KP3 out of external sync and see what the actual BPM was. It was consistently wrong. Even if I kept the master tempo locked, each time I took the KP3 in and out of external sync it had guessed an entirely different BPM, sometimes by several beats. Now mind you, I can't be sure this is actual evidence of a meandering, because to my ears all beat synced effects AND properly executed phrase sampling stayed in time as I described earlier. But as the previous poster indicated, there do seem to be quantifiable inaccuracies.
Auto BPM also has issues - you'll often be needing to tap tempos to help it along. And keeping recorded samples in sync over a period of time will require occasional 'align' corrections.
In my view, the KP3 sampler is basically a glorified "tape looper" unless you record one-shots. After reading (or should I say 'deciphering') the KP3 manual a bit more, it occurred to me that one workaround might be to record a loop as a one-shot (make your recorded loop length 16 bars but manually stop it at the end of 8 bars). and trigger it with a note-on message played by the master sequencer. This would of course require a dedicated track (and some pre-planning) and would essentially play a note-on message on beat one for a given pattern. It could potentially be a solution for those who want to trigger prerecorded loops that they load into the KP3 via the SD card.
This has been reported everywhere. Finally picking up steam. Many peeps staying away from this unit due to the midi bpm issues. It's not just auto BPM issue, it will drift quickly off of a master midi clock beat as well. Korg's answer to me was "use KP3 as the master and not the slave". Sorry, but this works fine in my KP2.
b.
b.
I believe that is part of the same issue. We're almost done fixing the bug - stay tuned!
Regards,
Jerry
Regards,
Jerry
Khazul wrote:jerrythek wrote:Hello folks:
Thanks for all your input regarding the Auto-BPM sensing. Korg Inc. has been looking at this based on our reporting your threads and we "think" we found a small bug in the smoothing of the BPM sensing mechanism. We're looking into it and will get back to you as soon as we have an answer. That's all I really know right now, but thanks to your input we're "on it".
Regards,
Jerry
Excellent news.
While you are at it, could you also look at the BPM smoothing that is applied to incoming midi sync, as Im finding that slaved it to an external midi clock seems to introduce alot of artifacts into the delays (ie odd squeaks as if the delay is sometimes having its time randomly modulated). The end result is its not really useable when clock slaved.
-
decrepitude
- Full Member
- Posts: 169
- Joined: Tue Dec 26, 2006 4:20 pm
- Contact:
-
MisterRider
- Posts: 18
- Joined: Wed Aug 06, 2008 2:36 pm
- Location: Earth
auto BPM and midi sync follow up
Soo... did this bug ever get worked out? No activity on the thread for quite some time
Re: auto BPM and midi sync follow up
I just downloaded OS 2.0 last week, and while the situation is far better than it was, I still have sync issues after a 4-bar sample loops more than 15-20 times. (I've got my KP3 sync'd up via MIDI to my EMX-1 Electribe, another Korg box, with the EMX clock set as master.)MisterRider wrote:Soo... did this bug ever get worked out? No activity on the thread for quite some time
That said, sweet that the company is at least working on the issue -- thanks, you Korg OS tweakers, for checking these forums. Glad there's at least one company out there that listens, unlike a few others I could mention.
EMX-1, KP3, MachineDrum UW+, Octatrack, FCB-1010 Pedal, Mackie PA, Taylor T5
http://dubathonic.bandcamp.com/
http://dubathonic.bandcamp.com/
-
MisterRider
- Posts: 18
- Joined: Wed Aug 06, 2008 2:36 pm
- Location: Earth
Yeah, I've had OS 2.0 for a while now, and I'm not sure KORG is working actively on any issues regarding sync - at least, I haven'[t seen any recent activity. My feeling is that it is not a feature they can fix easily without re-designing the basics of their sync engine, which apparently is based on MIDI clock *without* song position pointer. There is no way to accurately sync without using SPP and well, the KP3 doesn't. Understandable, as the device is primarily a DJ effect tool, and not a serious MIDI production tool (like the EMX for example).
IMHO, that sucks! I expected more from KORG when they advertised a proper MIDI sync. And by "more" I just mean "works..."
IMHO, that sucks! I expected more from KORG when they advertised a proper MIDI sync. And by "more" I just mean "works..."
-
MisterRider
- Posts: 18
- Joined: Wed Aug 06, 2008 2:36 pm
- Location: Earth
PS for Jerry
PS - Jerry! Any news on this fix? It's almost happy birthday time on the bug fix ...
jerrythek wrote:I believe that is part of the same issue. We're almost done fixing the bug - stay tuned!
Regards,
Jerry