Collaboration between hardware and software developers isn't common in the world of PC music — but this month Martin Walker reports on several attempts to bridge the divide.
Whenever a soundcard with lots of juicy DSP features is released, new owners become frustrated that their MIDI + Audio sequencer can't access them (or at least not until the next update appears, and sometimes not even then). In the interim, support may be limited to a third‑party utility, but while some of these are excellent in their own right, nothing quite compares to being able to control all your soundcard features from within a favourite sequencer. The problem is that there has never been a standard way to let software talk to hardware, which means that sequencer developers have to write new code every time a new soundcard is released.
AudioX is an open driver specification proposed by Cakewalk to overcome these current limitations, and fortunately it has already received wide support from other software and hardware companies. Back in December, Yamaha, Event, and Digital Audio Labs announced plans to support AudioX, and now Emagic, Steinberg, Lexicon, and the Frontier Design Group have also joined the group.
The Cakewalk solution is a cross‑platform one, working on both PCs and Macs and aiming to control all advanced aspects of soundcard control. These include SMPTE sync, varispeed, patching of software mixing channels to soundcard digital mixing channels and low‑latency peak metering, as well as adjustments to volume, pan, solo and mute controls. In effect, the main sequencer application will be able to directly access any relevant controls available on the soundcard itself in low‑latency digital form.
Just for once, what's proposed does seem to benefit all concerned — sequencer developers will find it much quicker and easier to support new soundcards (although they will still have to put in programming effort), and soundcard developers who include AudioX hooks in their drivers will make their particular features more easily available. In future, we may be able to buy a soundcard safe in the knowledge that if it has zero‑latency hardware monitoring, an on‑board digital mixer, or hardware‑based effects, we will be far more likely to eventually be able to access these directly from our sequencers, without having to switch to another program or set up convoluted routing through several software mixers.
Channelled Efforts
Steinberg and Propellerheads have also announced that their ReWire technology, for transferring audio between applications in real time, is to be opened up for use by other developers. ReWire provides sample‑accurate sync of up to 64 channels of real‑time audio, so that data can be transferred and controlled (using a single set of transport controls) between software applications. It has already been used in the latest version (2.0) of ReBirth to allow real‑time streaming of audio from ReBirth to Cubase VST. Given the current problems faced by people attempting to run VST and software samplers and synths simultaneously, being able to have all virtual music sources appearing as extra audio channels inside VST would solve a lot of problems. Let's hope that third‑party developers take up the offer.
Since Steinberg's ASIO standard is now being widely adopted by other developers, such as Emagic, it's good to hear that this streaming technology is also being extended. Steinberg have announced various new features that will be added to their next level ASIO (2.0), probably the most important of which is that it will support sample‑accurate sync with audio cards that have an ADAT Sync port, such as the Sonorus STUDI/O and Lexicon Studio.
The VST plug‑in format is also being upgraded to version 2.0, and this will finally allow MIDI control, as well as the integration of software synths into its real‑time audio processing (assuming that the software synth is being processed as a plug‑in). You may remember from my review of the Waldorf D‑Pole plug‑in (SOS November 1998) that its LFO syncing to audio was a bit of a fudge (Waldorf's own words), because VST didn't allow MIDI signals to trigger plug‑ins directly. VST 2.0 will remove this limitation.
Space Station Pro
Space Station Pro 1.4 is a software sampler for DOS or Windows 95/98 that uses system RAM for sample storage. It is designed to communicate directly with the DSP chip of the SoundBlaster 16, 32, or 64 soundcard range, although SB Live! drivers are also expected shortly. The benefit of this tight integration (as in the case of the original release of Seer Systems' Reality software synth — see SOS November 1997 for a review) is very low latency: Digital Audio Innovations claim a figure of 2.8mS. Also, unlike any SoundBlaster card, you can use as much of your RAM to store samples as you can spare; apparently up to 4Gb is theoretically supported!
SSPro is 16‑channel MIDI multitimbral, and up to a maximum of 128‑note stereo polyphony is supported, depending on the processor speed of the host computer. DAI claim 67 notes at 44.1kHz for a Pentium II 233MHz running under Windows 95 but, unusually, the software also caters for processors as slow as a 386, with reduced polyphony when running under DOS rather than Windows! A more typical Pentium 166MHz machine will, on average, achieve 48 voices at 44.1kHz under DOS. This excellent achievement is largely due to the fact that RAM is used for sample storage — SSPro doesn't access the hard drive at all when running, and this allows the speed of the processor to determine overall performance, which makes it far more efficient than Gigasampler, for instance (see review in SOS December 1998).
The software is comprehensive and is split into various 'zones', providing extensive sample treatment, including envelopes and LFOs. A comprehensive WAV file editor is also included. Unfortunately, SSPro can only be addressed as a MIDI device using the SoundBlaster external MIDI In socket or directly from SSPro itself — if you want to use it when simultaneously running a Windows sequencer, you need to loop a MIDI cable from a second MIDI Out to the In of the SoundBlaster card.
Although DOS applications rarely look as elegant as Windows ones, they are always more responsive, and SSPro already has an avid following among those with older PCs. Even those who have upgraded can give an old machine a longer and more useful life by turning it into a dedicated sampler. If you have a SoundBlaster soundcard lying dormant or, even better, an older PC with one inside, SSPro would seem a very good way to squeeze a lot more life out of it. Because the program runs entirely as a DOS application in RAM, it will also carry on running even if your hard drive blows up or Windows crashes, which will interest live performers. This is tightly written, performance‑driven software at its best.
Supplied on a single CD‑ROM that also contains over 1000 instrument samples in a wide variety of styles, SSPro 1.4 costs £117.50 including VAT, and a light version is also available with 500 instruments at a lower £58.75. You can obtain further details from the Digital Audio Innovations web site (www.digitalaudioinnovation.com)
Battle Of The Processors
Finally this month, I'm indebted to Kevin Perry for providing some very useful figures for his AMD K6‑2 400 processor in last month's review of Audio Architect — this has enabled me to produce the graph shown above, so that finally some direct correlations can be made between comparable AMD and Intel processors. I have conducted my tests using exactly the same software and settings as Kevin quotes — Sound Forge with both 16‑ and 24‑bit calculations, on mono 44.1kHz data. However, his comparison of the AMD 400MHz processor with my now sadly retired Pentium 166MHz MMX has been brought more up to date with the results I obtained from my newer Pentium II 300MHz processor.
Where possible, I've also shown the figures I measured with the Celeron 400MHz device that I borrowed from Red Submarine and used in this month's PC Musician feature (starting on page 182), although these tests were all carried out on stereo data, and sadly the Celeron had been returned before Kevin's test results appeared. Switching to stereo data makes little difference in the case of reverbs (among the most processor‑demanding of all effects) but the compression and dynamic effects (Waves C1 and L1) show a significant rise when processing two channels. I fully agree with Kevin that if you are mainly using MIDI, floating‑point performance doesn't really enter the picture — as I say in this month's PC Musician feature.
Most of the figures show low processor requirements, in the order of five percent or less for individual effects, and these are fairly even between the processors. However, the effect that always causes the most problems is reverb, since it takes far more floating‑point calculations, so the most telling result is that of the Waves TrueVerb (as always), where even my Pentium II 300MHz is twice as fast as the AMD K6‑2 running at 400MHz. The AMD K6‑2 400MHz is currently available for about £130 from PC suppliers, and the Celeron 400MHz for £135, so this seems a fairer comparison, and while the Celeron 400 takes 13.6 percent of its power to run the TrueVerb, according to Kevin's own figures the AMD K6‑2 400 takes 31 percent. This makes the Celeron 2.3 times faster.
Seek And Ye Shall Find
Here's a tiny tip that you may not have come across. The Windows Find Files utility is a useful way to track down recently saved data, or files containing specific words and phrases. However, you may not have noticed the 'Save Search' option in the File menu. If, like me, you regularly set up the same complex searches, this can save a lot of time. First, set up your search options, and then click on 'Save Search'. A new shortcut will appear on the desktop, and you can click on this in future to launch the Find Files window, with all options set up.
In The Pipeline
- Following my review of the new Creamware Pulsar (SOS March 1998), more information is now appearing about third‑party support. Creamware themselves refer to these companies as "pioneers", and they include Prosoniq, Quantec (room simulation), Sonic Timeworks (reverb), Spectral Design, C‑Mexx (remote control mixer), Wildcat Canyon Software (Autoscore pitch‑to‑MIDI software), and NemeSys (Gigasampler). The named companies have apparently stated that they have either started program development or intend to, and product launch dates may therefore be some months away.
- Seer Systems have announced a low‑cost software synth called SurReal. It features five different synthesis engines (sample playback, FM, physical modelling, analogue, and modal), and is 64‑note polyphonic and 16‑channel multitimbral. SoundFonts are also supported, and various effects, such as reverb, chorus, flanging, delay, and doubling, are provided. In essence, it is a cut‑down, preset version of Reality, although a 'Vari' knob allows some changes to be made to each voice. At a retail price of $129, SurReal should find a lot of potential purchasers.