valnar wrote:

I can also use a serial (RS-232C) to MIDI adapter which I have for my Roland, but then I would need a USB-to-serial adapter. Sounds cludgy, but maybe that is best? I dunno. Does it matter? Opinions?

For WinXP, if your motherboard does not have an actual RS-232c DB-9 serial port AND you already own a Roland SC-88 you might consider the following:

1.) Get a two-port serial DB-9 PCI card if you have the free PCI slots in your WinXP based PC workstation. The serial port to Roland SC-88 "Computer IN" connection has the lowest CPU overhead of all connection methods. It also has the lowest latency. Use a serial port with the proper serial cable in combination with the Yamaha CBX-MIDI v2.0 serial port driver and you are in business.

2.) If you go with a USB-MIDI interface then do not consider trying to use a USB-to-serial device and then connecting a serial-to-Roland MIDI cable. You suffer the USB protocol overhead plus the USB-to-serial conversion penalty and conversion anomolies. Just go ahead and get a Roland USB-to-MIDI device such as the UM-1EX and be done with it...

More Roland SC-88 MIDI/cable/driver details in this VOGONs post.

Hope this helps,

dvwjr

Thanks for the reply. So is there any benefit (or detriment) over a sound card with midi ports on it? What is the latency difference between that, serial and USB?

I ask because I see a few posts elsewhere that people point out the fact that the ASUS Xonar and M-Audio 24/96 comes with MIDI, like it's a good thing. They make a big deal about the MIDI bracket. So, why? If you have an external synthesizer, why not just use USB? Wouldn't that be the most futureproof?

valnar wrote:

Thanks for the reply. So is there any benefit (or detriment) over a sound card with midi ports on it? What is the latency difference between that, serial and USB?

The sound card with onboard current-loop MIDI would probably be the most expensive, with the least latency. See this post about halfway down for more data on MIDI latency.

valnar wrote:

I ask because I see a few posts elsewhere that people point out the fact that the ASUS Xonar and M-Audio 24/96 comes with MIDI, like it's a good thing. They make a big deal about the MIDI bracket. So, why? If you have an external synthesizer, why not just use USB? Wouldn't that be the most futureproof?

USB is 'future-proof', however since USB MIDI uses asynchronous timing for MIDI's aperiodic data, unlike USB audo which uses isosynchronous timing (which helps to guarentee delivery time) it suffers from greater MIDI latency/jitter. For modern serial protocol products, Firewire-MIDI has lower latency and jitter than USB-MIDI.

The best MIDI latency/jitter probably comes from either a PCI card with a true MIDI port, or for older ISA systems a SoundBlaster with its MIDI/Joystick port. Next would be a true serial port for Roland/Yamaha MIDI devices with serial port MINI-DIN 'computer' connectors. Firewire products next, with proprietary protocol Roland/Edirol USB-MIDI products next, with generic USB-MIDI units last.

Hope this helps,

dvwjr

Is there a midi-to-firewire cable you recommend? I have firewire.

valnar wrote:

Is there a midi-to-firewire cable you recommend? I have firewire.

No Firewire/MIDI recommendations since I do not have one. Most of the commerical Firewire/MIDI products just have the MIDI port as an add-on to Fire-wire/Audio external boxes with connections for various audio inputs. Ususally runs in the $250-$800 range... For use with Dosbox and MIDI output stick to PCI/MIDI or PCI/Serial for WinXP. Cheapest and least latency/jitter...

dvwjr

Do you know of any program or method to test the latency of the various output interfaces?

Most apps like Cubase, Sonar and the like have latency testing built into their driver selection portions. However, the latency they generally test for is INPUT from a controller (for use in driving virtual synths).

There are more factors at play that have an impact on latency, than the midi interface of choice alone.

Here's a good thread on the subject of latency and how to test for it in its many guises:-

http://forums.musicplayer.com/ubbthreads.php/ … 9892159003bad93

Also (referenced from same thread) check that your drivers and/or apps are doing DPC (Deferred Procedure Calls) properly and are not the direct cause of latency themselves. http://www.thesycon.de/deu/latency_check.shtml

valnar wrote:

Do you know of any program or method to test the latency of the various output interfaces?

Here is a MIDI test application from Earth Vega Connection Audio software. The name of the attached file is MIDITEST46.ZIP and it contains the Win32 executable MidiTest.exe version 4.6.231 date Feb-07-2007.

This should get your interface testing/comparisons in the MIDI latency/jitter ballpark, can not say how accurate it really is as compared to commercial MIDI audio apps... The MIDITEST internal time-stamping does use the CPU RDTSC (read time stamp counter) rather than just the Win32 MM system timer (~1ms resolution) so at least the timer has more resolution than what is being measured.

I have used MIDITEST in 2006 to compare a Roland USB-MIDI solution to a Roland/Yamaha serial port MIDI drivers for my SC-88VL/MT-32 chain. Went with the Intel motherboard serial port/DB-9/Mini-DIN with the Yamaha CBX version 2.0 WinXP driver instead.

Hope this helps,

dvwjr

Cool, thanks. I'll give it a shot.

Have a look at this page:

http://www.soundonsound.com/sos/Oct04/articles/qa1004-7.htm

It confirms what dvwjr said: PCI cards are the fastest for MIDI interfacing.

My PCI-based MIDI interface gave me around 3.6ms input latency when capturing a keyboard performance, with a latency jitter of just 0.2ms; a serial-port interface gave around 4.2ms with jitter of 1.2ms; and a USB interface gave about 4.8ms with jitter of up to 1.9ms, all running under Windows XP

The problem is, where can someone find a PCI-based MIDI interface card nowadays?