Having explained some of the audio and MIDI interfacing problems facing owners of new‑style Macs, Paul Wiffen looks at practicalities, testing one USB MIDI and one USB MIDI + Audio interface to see what pitfalls await the unwary... This is the third article in a four‑part series.
As I pointed out last month, the situation regarding MIDI and audio interfacing for new‑style Macs is changing on a weekly basis at present, making it necessary when writing this ongoing series to spend time each month recapping on developments that have occurred between the appearance of each copy of SOS. This month I am again happy to report a healthy response from readers to the previous instalment of this series — I've summarised some of your problems and solutions in the 'Other Voices' box on page 78. I must say that, with the exception of a couple of store salesmen who bemoaned the fact that we have let a few cats out of the bag, the feedback has all been very positive. These salesmen claimed that their customers never saw any of the problems which I had mentioned, because they knew all the solutions to them and so pre‑empted them. Why was I worrying their customers unnecessarily? My response was that they should be using this knowledge of problems and solutions to attract customers rather then make out that everything in the Mac garden is rosy. My advice to you is to use some of the problems that I have flagged‑up to test the knowledge of the stores you are speaking to! Ask if the Digigram VXPocket is suitable for use with softsynths, and if they say "yes", ask if the latency problems (see last month) have been fixed. Ask how you can use an old‑style serial port on a new Mac, and see if they mention Geethree's Stealth port, Griffin Technology's gPort or the Romulus card from MegaWolf, or if they go straight to trying to sell you a USB device.
As you'll remember from the end of last month's piece, I planned this issue to look at some of the most recent hardware additions to the USB fray, in particular those interfaces which combine MIDI and digital audio I/O in one piece of hardware. Unfortunately, I had to restrict the scope of my investigations, as some of the hardware I hoped to take a look at (such as Event's EZBus) is still not shipping in the UK at the time of writing. Other devices are available, but not yet compatible with the Macintosh, only Windows PCs.
Into this bracket fall Roland's UA100 MIDI + audio interface and the U8, a combination MIDI + audio interface and control surface (see the reviews of these devices on the PC in SOS February and June 2000 respectively). Neither of these can yet be used reliably with Mac OS 9, due to the see‑saw nature of USB support under the Apple Sound Manager. Speaking to the head of Edirol (Roland's European computer products division) at the recent Music Live 2000 show in Birmingham, I learnt that they are still only targeting PC customers with these units. I brought up the Propagamma generic USB driver I mentioned last month in the hope that this might offer a way around the Sound Manager support problem, but Roland apparently remain concerned about the tech support difficulties involved when a third‑party driver is used with a product. Rather than get involved in the kind of buck‑passing that can develop from this type of situation, they are still hoping that Mac OS X will prove to be the solution to all the problems Mac owners are experiencing with USB. This would allow them to make both the U8 and the UA100 available to Mac owners. At the moment, however, Edirol only recommend the UA30 stereo audio‑only interface with the Mac (and then only with Mac OS 9.0.4). As a result, this was the only one we requested for testing.
The results I obtained were very similar to those Vic Lennard did recently (see Crosstalk, SOS November 2000). In other words, I made perfectly good recordings via the UA30 but there was some instability and glitching when I attempted to play those recordings back. Not surprisingly, I still don't feel the UA30 can be recommended wholeheartedly to Mac users. Swissonic's Studio D USB line mixer/audio interface does not perform perfectly under Sound Manager control and needs the Propagamma ASIO driver to make it work reliably, so I continue to think the UA30 would benefit from a custom ASIO driver.
As a result of all this, the only combined MIDI and audio USB interface available for testing this month was the stylish Tascam USB428. It has eight faders, four audio inputs, two MIDI ports, a four‑band EQ section, and its transport controls are designed to let you operate your MIDI + audio sequencer remotely. In fact, despite the Tascam branding, the US428 was developed by US company Frontier Designs. This should come as no surprise to many of you, as Frontier are one of the few companies producing PCI products for the Mac which offer both MIDI and audio interfacing on one card (the Digi 001 being the only other one I am aware of).
The USB428 is without doubt a fantastic concept; it makes for easily the most compact home or portable setup when added to an iMac or a PowerBook running something like Cubase VST. Derek Johnson and Debbie Poyser will be giving it the full SOS review treatment on a G4 in a forthcoming issue. However, as it came for review while I was working on this series, it made sense for the Editorial team at SOS to pass it to me for investigation. The major question for me was whether the Universal Serial bus on a Mac really could reliably handle the US428's promised four channels of audio input, two channels of output, and two MIDI busses simultaneously...
Having run the installer CD on a PowerBook and rebooted, I went to Cubase VST/24's Audio System window and found drivers for both 16‑bit and 24‑bit operation installed. This is quite remarkable in a unit costing less than £500 (the US428 retails for £449). The latency Cubase reported was 26 milliseconds, 4mS less than if using the standard Sound Manager ASIO driver for both the 16 and 24‑bit drivers. I decided to walk before I ran, and selected the 16‑bit driver. I put four tracks into record at once while monitoring and found that all were recorded perfectly with no glitches or clicks. This was a good start (and something my sources in the industry had been very scept ical about), but there was a long way to go yet. Would it manage this sort of reliable performance while triggering 32 MIDI channels down the same USB cable? It was time to look at getting the MIDI side of things working.
Sadly, this was where this started to go wrong. The first time I ran the OMS Studio Setup procedure only a USB428 Control Port numbered '3' appeared. This presumably facilitates the control of VST's internal faders, as when I selected test mode and moved a fader on the Tascam I got a 'MIDI Received' message from OMS. However, Ports 1 and 2 were nowhere to be seen. I tried plugging a keyboard into the first port, and it was clearly working, because the green MIDI In LED on the front panel of the USB428 flashed. I ran the MIDI Cards and Interfaces operation again in case it could now see something on those inputs, but there were still no Ports 1 & 2. The US428 manual isn't much use, as it talks almost exclusively about the copy of Steinberg's Cubasis for the PC which is supplied free with the unit. The only mention of Macintosh troubleshooting concerned the choice of 16 or 24‑bit audio drivers, which (naturally) I had installed first time with no problems at all!
As is so often the way with bizarre faults like this, after three or four more attempts, Ports 1 & 2 suddenly appeared as if by magic, without me being at all clear what I had done to make this happen. Thinking my luck must be changing, I pressed on, trying to get the remote operation of Cubase going via the USB Control Port which had been there all along (so I could try moving faders as well whilst playing back MIDI and recording). I was able to select this port for both Input and Output under VST Remote, but there was no USB428 protocol in the list of options in my copy of Cubase VST v4.1r2, the latest version for the Mac at the time of writing. I could see from the manual that on the bundled PC version of Cubasis the USB428 option appears automatically, but maybe this needs to be implemented in the full version of Cubase VST before this is possible.
Giving up on remote control, I went back to ordinary MIDI sequencing, set up a MIDI click to the keyboard to play along to, and then recorded a drum part. To be on the safe side, I quantised the part before playing, it because I didn't want any vagaries of human performance to cloud my assessment of the tightness of the timing. I had planned to record at least a couple of dozen tracks on different MIDI channels to really give it a good test, but sadly, I never got that far. Even playing back the one drum track, the timing was all over the place; even that of the MIDI click.
I had to conclude that the MIDI side of the USB428 was falling at the first hurdle I could put in front of it. I didn't go on to test the audio and MIDI simultaneously — there seemed little point when the MIDI was performing so badly on its own. I suspect the unreliable OMS/USB combination is at the heart of this (see the first and second instalments of this series for a full explanation of this). Consequently, I reckon that direct USB access by the sequencer would cure the problem, just as it seems to with Emagic and MOTU software/USB hardware pairings. Apparently, a special Mac version of Cubasis for the USB 428 is in preparation, so this timing inconsistency may be addressed fairly soon (I do hope they put the direct driver for the US428 in the more upmarket VST packages as well!). Until then, based on my experiences, I cannot recommend the 428 for MIDI interfacing on the Mac at all at present. I will be reporting back on any developments that follow (see below), but for the moment, it looks like the combination of MIDI and digital audio in a single USB device isn't a workable reality for the owners of new Macs.
Just as we were going to press, a MOTU Digital Performer driver for the USB428 was announced. Unfortunately, email problems prevented me from receiving it in time to evaluate it for this article, so I will attempt to report back on its operation next month.
Just like last month, all the information in this article should be considered correct at the time of going to press, but subject to rapid change. I'll report back next month with any updates to the USB scene, and after that will continue to inform everyone via SOS's Apple Notes column, which I will be taking over next month, not this month as I reported in last month's SOS (I have to let Vic Lennard write his parting column, after all!). If you have any further comments to make or updates to bring to my attention, please mail them to the SOS office, or email me via the new Apple Notes address: email@example.com.
Many thanks to this month's crop of readers who got back to us with their personal experiences of audio on the new Macs. You may remember from last month that although Digigram's VXPocket PCMCIA I/O card provides excellent audio quality, its latency problems make it unsuitable for multitracking. Mark Bromwich emailed in to report that he also suffers from occasional clicks and dropouts when using Cycling '74's MSP through the VXPocket and so far Digigram have been unable to suggest a fix for this. He does applaud their efforts to solve his problem though, as they have sent him several beta versions of their drivers using smaller buffer sizes. Sadly, these did not work properly either, although I am interested to hear that the company are trying smaller buffer sizes — it sounds like they may be trying to reduce the input latency which makes their card so difficult to use when overdubbing. Mark explains, "The MSP patches I am working with are only using 30 percent of my available CPU power, so I would have thought the card would work seamlessly. This and the poor latency has made me go back to using my Mac's built‑in convertors, which is not an ideal solution. I am not alone in this, as people on the IRCAM Internet forum have also experienced similar problems."
I suggested last month that keeping USB MIDI and audio devices on different USB busses might prevent them stealing bandwidth from each other. On reading this, John Krogh of California was prompted to remind me that whilst this may well be a valid approach on G4s, which have separate busses on each USB connector, this does not apply to blue‑and‑white G3s. Although these Macs have two USB connectors, John has it on Apple's authority that both ports are on the same buss. I remember being told something similar unofficially at one stage, but it is nice to have it confirmed!
Finally, Thom Cowland emailed to ask if I had seen the FireWire‑based Metric Halo Mobile I/O which he thought looked like a perfect solution for analogue I/O‑less PowerBook and Cube users. In fact, I included it in the final part of my mLAN series (see SOS August‑November 2000), for completeness, as one of the new developments at AES, and you can read more about it in the SOS News pages this month (see page 7). In fact, the Mobile I/O doesn't actually connect via mLAN but a proprietary FireWire protocol. At AES, Metric Halo were quoting delivery in March 2001, and the unit on display was certainly not a working model. However, look out for review coverage in SOS as soon as it is shipping.
Setting the right DAT recording level: SOS January 1995. Noise and how to avoid it: SOS May 1995. A Concise Guide to Compression & Limiting: SOS April 1996. The Mysteries of Metering: SOS May 1996.Minimising Mixer and Effects Noise: SOS July 1996.