Subject: Re: [AUDITORY] MIDI Drum Pads: Delay and temporal resolution From: "Mikael.Fernstrom" <Mikael.Fernstrom@xxxxxxxx> Date: Wed, 19 Jun 2013 10:49:07 +0100 List-Archive:<http://lists.mcgill.ca/scripts/wa.exe?LIST=AUDITORY>--Apple-Mail-9F5CCFD4-C81D-4CCC-BBB9-B093571B2413 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=us-ascii All things MIDI has ms latency. A less expensive data acquisition system is an Arduino with some simple inte= rface wiring. Cost around 25 euro. /Mikael Sent from my iPad On 19 Jun 2013, at 09:37, "Julius Smith" <jos@xxxxxxxx> wrote: > Apparently one can get a tap-to-sound latency of 5.8 ms in iOS: >=20 > http://www.synthtopia.com/content/2011/12/03/developer-it-is-not-possible-= to-play-music-on-an-android-phone/ >=20 > To get below 1 ms I would probably connect a piezo film to an Arduino or s= omething like that. >=20 > - Julius >=20 > At 10:04 PM 6/18/2013, SKoT McDonald wrote: >> In 2006 at FXpansion, we made measurements of different drum module's lat= encies as part of an investigation into why certain users were experiencing b= ad latency with their e-drum kits when using our BFD range of drum software.= >>=20 >> We found that Roland TD-20s were quite slow - it seems their propitiatory= COSM modelling of impacts added 1-2 ms, even for MIDI-only event output; po= sitional information was sent before the note onset in the form of MIDI CC m= essages - each 3-byte MIDI message adds 1ms - and some pads were sending 2-3= MIDI CC messages (hihats in particular!). The TD20 of the time had no USB M= IDI option. >>=20 >> A Yamaha DTX kit by contrast had less pre-processing of impacts, a USB co= nnection for MIDI, and less MIDI CC positional information, which I think yo= u could switch off too - this made the Yamaha "brains" about 3-4 ms faster t= o get a MIDI Note to the sequencer or instrument on your computer. >>=20 >> Our audio devices tended to be high end, such as RME Firefaces with 64 sa= mple buffers (~1.5ms @xxxxxxxx 44.1kHz). >>=20 >> Considering that ~6ms is the border of average human timing discernment, w= e found "rhythmically highly trained" subjects, like drummers, were capable o= f being put off by the TD-20's lag, despite being the highest priced kit in t= he market. We hassled Roland for many years to at least add a MIDI USB conne= ction... >>=20 >> Cheers, >>=20 >> SKoT McDonald >>=20 >>=20 >>=20 >> On 18/06/2013 7:09 PM, Andre Holzapfel wrote: >>> Dear Benjamin, >>>=20 >>> we will have a paper in this years SMC conference related to the acquisi= tion of tapping data using software implemented at INESC TEC in Porto. >>>=20 >>> We did an experiment to determine the overall latency when using a MIDI i= nterface (Roland Handsonic HPD-10) connected through a Digidesign USB sound c= ard to a 2012 iMac. The obtained latency was 17ms. I do not know how much yo= u can gain by using more high end interfaces, but I would agree with the pre= vious response, that it is very unlikely to get even close to sub-millisecon= d accuracy using MIDI. >>>=20 >>> Please let me know about your further investigations, contact me if you l= ike to obtain the paper in advance. >>>=20 >>> Best, >>>=20 >>> Andre Holzapfel >>>=20 >>>=20 >>> Quoting Benjamin Schultz <benjamin.schultz@xxxxxxxx>: >>>=20 >>>> Dear Auditory List, >>>>=20 >>>> I am currently looking at purchasing a MIDI Drum Pad for experimental r= esearch involving finger tapping. There appears to be very little informatio= n regarding how much of a time delay there is between the pads being tapped a= nd the sound being produced (via external MIDI devices or the drum pad's int= ernal sequencer). >>>>=20 >>>> I'm looking to have the smallest time delay possible (sub-millisecond w= ould be ideal). Does anyone have any advice about the best device to use and= what kinds of time delays I should expect from the device? Their unit of pu= lses per quarter note does not appear to be very informative. >>>>=20 >>>> Kind regards, >>>>=20 >>>> Dr. Benjamin G. Schultz >>>> Post-doctoral Fellow >>>>=20 >>>> Sequence Production Lab >>>> McGill University >>>> Stewart Biology Building >>>> 1205 Dr. Penfield Ave., Montreal H3A 1B1 >>>> Montreal, Quebec H3A 1B1 >>>> Canada >>>> Tel.: 514-398-5270 >>>> Fax: 514-398-4896 >>>>=20 >>>> Email: benjamin.schultz@xxxxxxxx >>>> http://www.mcgill.ca/spl/members/currentmembers#SCHULTZ < http://www.mc= gill.ca/spl/members/currentmembers%23SCHULTZ>=20 >>>>=20 >>>=20 >>>=20 >>>=20 >>> ---------------------------------------------------------------- >>> This message was sent using IMP, the Internet Messaging Program. >=20 >=20 > Julius O. Smith III <jos@xxxxxxxx> > Prof. of Music and Assoc. Prof. (by courtesy) of Electrical Engineering > CCRMA, Stanford University > http://ccrma.stanford.edu/~jos/ --Apple-Mail-9F5CCFD4-C81D-4CCC-BBB9-B093571B2413 Content-Transfer-Encoding: 7bit Content-Type: text/html; charset=utf-8 <html><head></head><body bgcolor="#FFFFFF"><div>All things MIDI has ms latency.</div><div>A less expensive data acquisition system is an Arduino with some simple interface wiring.</div><div>Cost around 25 euro.<br><br><div>/Mikael</div>Sent from my iPad</div><div><br>On 19 Jun 2013, at 09:37, "Julius Smith" <<a href="mailto:jos@xxxxxxxx">jos@xxxxxxxx</a>> wrote:<br><br></div><div></div><blockquote type="cite"><div> <font size="3">Apparently one can get a tap-to-sound latency of 5.8 ms in iOS:<br><br> <a href="http://www.synthtopia.com/content/2011/12/03/developer-it-is-not-possible-to-play-music-on-an-android-phone/" eudora="autourl"> http://www.synthtopia.com/content/2011/12/03/developer-it-is-not-possible-to-play-music-on-an-android-phone/<br> <br> </a>To get below 1 ms I would probably connect a piezo film to an Arduino or something like that.<br><br> - Julius<br><br> At 10:04 PM 6/18/2013, SKoT McDonald wrote:<br> <blockquote type="cite" class="cite" cite="">In 2006 at FXpansion, we made measurements of different drum module's latencies as part of an investigation into why certain users were experiencing bad latency with their e-drum kits when using our BFD range of drum software.<br><br> We found that Roland TD-20s were quite slow - it seems their propitiatory COSM modelling of impacts added 1-2 ms, even for MIDI-only event output; positional information was sent before the note onset in the form of MIDI CC messages - each 3-byte MIDI message adds 1ms - and some pads were sending 2-3 MIDI CC messages (hihats in particular!). The TD20 of the time had no USB MIDI option.<br><br> A Yamaha DTX kit by contrast had less pre-processing of impacts, a USB connection for MIDI, and less MIDI CC positional information, which I think you could switch off too - this made the Yamaha "brains" about 3-4 ms faster to get a MIDI Note to the sequencer or instrument on your computer.<br><br> Our audio devices tended to be high end, such as RME Firefaces with 64 sample buffers (~1.5ms @xxxxxxxx 44.1kHz).<br><br> Considering that ~6ms is the border of average human timing discernment, we found "rhythmically highly trained" subjects, like drummers, were capable of being put off by the TD-20's lag, despite being the highest priced kit in the market. We hassled Roland for many years to at least add a MIDI USB connection...<br><br> Cheers,<br><br> SKoT McDonald<br><br> <br><br> On 18/06/2013 7:09 PM, Andre Holzapfel wrote:<br> <blockquote type="cite" class="cite" cite="">Dear Benjamin,<br><br> we will have a paper in this years SMC conference related to the acquisition of tapping data using software implemented at INESC TEC in Porto.<br><br> We did an experiment to determine the overall latency when using a MIDI interface (Roland Handsonic HPD-10) connected through a Digidesign USB sound card to a 2012 iMac. The obtained latency was 17ms. I do not know how much you can gain by using more high end interfaces, but I would agree with the previous response, that it is very unlikely to get even close to sub-millisecond accuracy using MIDI.<br><br> Please let me know about your further investigations, contact me if you like to obtain the paper in advance.<br><br> Best,<br><br> Andre Holzapfel<br><br> <br> Quoting Benjamin Schultz <<a href="mailto:benjamin.schultz@xxxxxxxx">benjamin.schultz@xxxxxxxx</a>>:<br><br> <blockquote type="cite" class="cite" cite="">Dear Auditory List,<br><br> I am currently looking at purchasing a MIDI Drum Pad for experimental research involving finger tapping. There appears to be very little information regarding how much of a time delay there is between the pads being tapped and the sound being produced (via external MIDI devices or the drum pad's internal sequencer).<br><br> I'm looking to have the smallest time delay possible (sub-millisecond would be ideal). Does anyone have any advice about the best device to use and what kinds of time delays I should expect from the device? Their unit of pulses per quarter note does not appear to be very informative.<br><br> Kind regards,<br><br> Dr. Benjamin G. Schultz<br> Post-doctoral Fellow<br><br> Sequence Production Lab<br> McGill University<br> Stewart Biology Building<br> 1205 Dr. Penfield Ave., Montreal H3A 1B1<br> Montreal, Quebec H3A 1B1<br> Canada<br> Tel.: 514-398-5270<br> Fax: 514-398-4896<br><br> Email: <a href="mailto:benjamin.schultz@xxxxxxxx">benjamin.schultz@xxxxxxxx</a><br> <a href="http://www.mcgill.ca/spl/members/currentmembers#SCHULTZ" eudora="autourl"> http://www.mcgill.ca/spl/members/currentmembers#SCHULTZ</a> <<a href="http://www.mcgill.ca/spl/members/currentmembers#SCHULTZ" eudora="autourl"> http://www.mcgill.ca/spl/members/currentmembers%23SCHULTZ</a>> <br><br> </blockquote><br><br> <br> ----------------------------------------------------------------<br> This message was sent using IMP, the Internet Messaging Program.</blockquote></blockquote> <x-sigsep></x-sigsep></font><p><font size="3"> <br> Julius O. Smith III <<a href="mailto:jos@xxxxxxxx">jos@xxxxxxxx</a>><br> Prof. of Music and Assoc. Prof. (by courtesy) of Electrical Engineering<br> CCRMA, Stanford University<br> <a href="http://ccrma.stanford.edu/~jos/" eudora="autourl"> http://ccrma.stanford.edu/~jos/</a> </font> </p></div></blockquote></body></html> --Apple-Mail-9F5CCFD4-C81D-4CCC-BBB9-B093571B2413--