SONAR – New Technology, More Power

PLEASE NOTE: This article has been archived. It first appeared on ProRec.com in June 2001, contributed by then Editor-in-Chief Rip Rowan. We will not be making any updates to the article. Please visit the home page for our latest content. Thank you!

One of the most exciting events in audio this year is the eagerly anticipated release of SONAR, Cakewalk’s replacement for the Pro Audio line of applications. Following in the footsteps of Pro Audio 9, SONAR is an integrated MIDI sequencer and audio multitrack recorder. However, SONAR has a number of new features and technologies which place it at the forefront of MIDI and audio technology.

Let’s face it: Pro Audio was approaching a turning point. After nine versions Pro Audio was saddled with a long feature list including many features that most people would consider esoteric (such as a programming language, CAL, and object-oriented studio controller builder, StudioWare) and some features which should be obsolete (such as destructive editing tools for lowering volume and normalizing) it was time to clean out the deadwood and make room for new growth. We will see, however, that more pruning could have been done.

But first, let’s review the new stuff. There’s a lot to cover.

Serious New Technology

To me the most impressive aspect of SONAR is that it is one of the first multitrack audio applications to truly capitalize on new technology from Microsoft: Windows 2000 and the WDM driver model. Besides offering increased OS stability, Windows 2000 supports symmetric multiprocessing, allowing a properly-written application to spread its resource requirements across multiple processors. SONAR takes full advantage of SMP, running strongly with two processors.

Some people will point out that having multiple processors does not actually double performance. Typically users will see, for example, about a 1.5X increase in CPU performance (meaning 50% more plug-ins) when running two processors. However, a side benefit is that even though certain application threads (like mixing) may bottleneck one CPU, the other CPU is usually not bottlenecked, providing, for example, fast user interface response even under heavy mixing loads. The bottom line is that with two or more processors, many functions in SONAR feel fast even when the app is loaded down.

WDM is Microsoft’s new driver model which provides kernel-level streaming. WDM drivers differ from MME drivers by having fewer program layers between the application and the hardware. The net result is that a properly-written WDM driver can achieve latencies as low as 1.5 ms – low enough to support realtime processing like software synthesizers and live audio effects monitoring. Look out ASIO, here comes WDM.

Why didn’t Cakewalk choose ASIO instead of WDM? Well, first there’s the obvious answer: Steinberg controls the ASIO spec, and could change the driver model without notice, thus keeping Cakewalk on uneven footing. However, there’s a more important answer for the consumer. WDM is an open driver model, and it’s endorsed by Microsoft.

Ah, Microsoft. As much as we all like to hate Microsoft, the fact remains that operating system support is the most critical factor in determining the success of a software platform. And few are the cases in which Microsoft technology did not dominate its competitors. By choosing WDM instead of ASIO, Cakewalk and Microsoft have created a powerful alliance. Moreover, other software companies are getting on board Microsoft’s WDM model, clearly paving the way for a truly professional audio driver standard for Windows. This is a Good Thing.

Note that you do not have to use Windows 2000 (or Windows XP) or WDM drivers in order to use SONAR. However, there are two aspects of SONAR that require low latency in order to be useful: live performance of software synthesizers and live audio effects monitoring. These features do not perform fast enough without WDM to be useful. You can make the move to SONAR when you’re ready, but to take full advantage of everything SONAR has to offer, you’ll want WDM, and to get the best performance, Windows 2000.

OK, enough of the under-the-hood stuff. Let’s listen to the bells and whistles.

Software Synths

Probably the hottest aspect of SONAR is its support for DXi software synthesizers. DXi – which stands for DirectX instrument – is a new standard for creating a software synthesizer that functions like a DirectX plugin. Functionally, DXi plugins work like Cubase vSt plugins – you just insert them like a plugin into the application and play them with any MIDI controller. Although Cakewalk created the DXi standard, it is based on standard DirectX technology, and, most importantly, is endorsed by Microsoft.

If you’ve never used a DXi or vSt synth before, then you may be missing the most important part of the technology: since the synth is plugged into a project file, its settings are saved with the project whenever you save, and are instantly recalled when you open the project. So instead of having a rack of hardware synths, which all have to be set to their appropriate settings in order to re-create a MIDI song, the synths all instantly recall their settings whenever you load the song. Softsynth plugins are great. I love being able to create a sound in a plugin, and follow it with a DX audio effect – say, a Rhodes piano synth plugged into an FX2 amp simulator plugged into an FX3 room simulator. Great sound, and emminently playable.

Of course, the other thing is simply that software synths are cool and sound great. SONAR comes bundled with several DXi instruments to help you get started, including a software version of the Roland Sound Canvas called VSC, a software Soundfont player called LiveSynth, a modular synth called Tassman, and Audio Simulation’s DreamStation. Tassman in particular is complex enough to justify its own article, and as a modular synth is exceeded in capability only by Native Instruments’ Reaktor synthesizer (a DXi version of Reaktor is also in the works). Suffice to say, if you’re a synth tweakhead, you’ll love Tassman. Between VSC, LiveSynth, DreamStation and Tassman, there’s more synthesis included in SONAR than you’re likely to find in your local music store.

And if the included and upcoming DXi synths aren’t enough for you, then check out Amulet’s VST Adapter 3.0. VST Adapter 3.0 allows any VST plugin to be used inside a DX application like SONAR, instantly giving the SONAR user access to any VST instrument or audio plugin. With the immediate ability to use any VST or DXi synth or audio plugin, SONAR now sets a new standard for offering more plugin power than any application currently available.

The other benefit to DXi technology is that it allows an audio plugin to be MIDI controlled. This allows you to automate the parameters of your EQs, reverbs, flangers, and vocoders. DX plugins must be rewritten to support this new feature, but SONAR ships with several plugins from PowerFX which already support MIDI automation. VST plugins running in SONAR using VST Adapter 3.0 can also be automated, just like they can in Cubase. However, DirectX audio effect parameters can be envelope automated at a resolution and accuracy much higher than MIDI-controlled plugins like vSt plugins. DX plugins must be rewritten to support this new feature, but SONAR ships with several effects from Power Technologies which already support automation.

Audio Looping

The other SONAR feature that has users talking is its support for Acidized audio loop files. This allows you to construct and arrange songs using loops, just like in Acid. SONAR takes this one step further by supporting MIDI loops which work just like audio loops. And, whereas Acid forces the user to a one-track-per-loop model, SONAR allows the user to mix loops on a single audio track.

The obvious question for any loop-based musician looking for MIDI support is, “which is better, Acid Pro 3.0 or SONAR?” This depends on the degree of MIDI support you need. If you just want to be able to embed some MIDI into your loop-based music, and you already use Acid and don’t use Cakewalk products, you’re probably best off sticking with Acid. However, if you want the ultimate combination of audio looping, a full-blown MIDI sequencer with integrated software synths, and excellent audio mixing, nothing on the market today can touch SONAR.

New User Interface

Responding to complaints about Pro Audio’s antiquated spreadsheet-style track view, SONAR features an entirely redesigned main user interface. This new track view shares a lot in common with Samplitude and Vegas, with graphical controls for both audio and MIDI tracks instead of the spreadsheet numbers. Users accustomed to the spreadsheet numbers will feel out of place at first, but the new user interface doesn’t take long to master. In no time I was making full use of the controls and getting work done quickly. Note, however, that the graphical nature of the user interface makes it more difficult to drive the application with keyboard controls. You’re better off driving with the mouse.

As you poke around in the track view you’ll quickly notice a few cool new features. SONAR features excellent metering capability on both inputs and outputs, with both peak and RMS hold and a variety of scaling options. When SONAR is running in its lowest latencies, these meters are extremely accurate and very fast – better than the metering available in any other multitrack application. Thanks to SONAR’s excellent multithreading, the meters also consume a paucity of system resources, so you’re safe using them at all times.

Another feature you’re likely to discover is SONAR’s improved envelope mixing capabilities. Every track send, including gain, volume, pan, and aux sends and pans can be controlled and automated via envelope, as can DirectX effects and DXi instruments. If you aren’t used to mixing with envelopes, you should be. In the right hands, a well-implemented envelope mixing tool can actually be more productive than a hardware mixing surface. SONAR also provides, for the first time in an audio application, envelope control over each Main and Aux bus. You can completely mix your song with envelopes, and never touch a fader.

Other Cool Stuff

A full feature rundown of SONAR would take too long to read. Important new features not already mentioned include 960 PPQM MIDI support for enhanced MIDI timing, automatic sync to any video source, slip editing, auto-crossfading on overlapped audio, phase switches on each audio channel, and strong dual monitor support. Most of the professional features which Cakewalk users have been wanting for years are now included in SONAR. Almost all capabilities of the Pro Audio platform have been retained, including video support, MIDI plugins, views for music notation and lyrics, and support for plug-in two-track editors like Soundforge and Cool Edit.

Performance

What really impressed me most about SONAR is its performance, which I found on par or better than the performance in Pro Audio 9.

I performed some benchmarks of the application using my trusty Roll Your Own computer: Celeron 533 overclocked to 800 MHz, 40 GB 7200 RPM IDE audio drives, 128 MB of RAM, and my MOTU 1224 soundcard. With this setup I could achieve up to 72 tracks of 24-bit / 44.1 KHz audio with rock-bottom latency: 2.9 ms! Bumping up to 35 ms latency took me to 110 tracks of 24/44.1 audio – low enough latency for very responsive mixing. For basic audio mixing capability, SONAR running on Windows 2000 using WDM drivers is able to perform as well or better than Pro Audio 9.

Softsynths are more demanding of the CPU, but for the most part are also well-behaved. I was able to run a number of LiveSynth plugins along with a dozen or so audio tracks. With a faster computer, I think that SONAR would be able to achieve the goal of near-zero latency even when playing dozens of synths, audio tracks, and audio plugins. As I stacked up audio tracks and software synths I found myself wishing for a little more horsepower, particularly when it came time to mix down. In particular I’m looking forward to a dual-processor board which should be able to offer much better SONAR performance than a single processor machine like mine. However, at the current time, the serious synthesist/recordist will want to maintain a separate machine dedicated to softsynths and samplers. I find that two “pretty good” computers offers better price / performance and stability for synthesis and recording than one “hoss box.”

When you first run SONAR it will “profile” your soundcard, finding and setting the optimum buffers for your hardware. It is important to remember that if you change your audio driver or its settings, you must let SONAR re-profile your soundcard. Additionally, SONAR provides a number of parameters for tweaking the audio engine – an advanced feature we really appreciate. While I tweaked with the various parameters, I found that, on my machine, the “out of the box” settings worked best overall. In the end the only change to the default configuration that I made was to turn the latency slider down to 2.9 ms and leave it there.

Stability and Reliability

For my tests, I wanted to try out the whole banana: SONAR running under Windows 2000 using WDM drivers. Knowing I was making the move from Windows 98 to Windows 2000, I planned some downtime in my schedule to make the switch, get my software reinstalled, and debug the environment.

I found the biggest challenge to be driver support for WDM and Windows 2000. My MIDI adapter was practically antique: an Opcode MIDI Translator. Out it went, replaced by a Midiman USB MIDI adapter. Also I needed three different versions of MOTU’s beta WDM drivers before my MOTU 1224 interface started working correctly with the operating system. Then there was the learning curve of getting Windows 2000 optimized for audio, which took a day to get most of the wrinkles ironed out. Finally, there was the setup of SONAR, to get it working in the manner to which I am accustomed.

I have found SONAR to be generally stable for most basic audio and MIDI tasks. In particular, audio recording and mixdown seems quite stable. The first week I had my gear up and running I used SONAR to record a live 2 hour concert using a 16-track MOTU setup. I recorded and mixed the concert without a single strange dropout, glitch, or crash. As a replacement for Pro Audio 9, SONAR did not slow me down at all once I had made the OS transition.

The new advanced features are much more prone to gotchas. In particular the software synths are prone to glitches and crashes. I found that about a dozen different Soundfonts were capable of crashing LiveSynth. Of course, this is technically a LiveSynth problem, not a SONAR problem, but the net result was that both SONAR and LiveSynth conspired to crash together, requiring a reboot and causing lost work. Tassman was also capable of crashing the application fairly regularly. In addition I found a number of little glitches and strange things in features like audio looping, file import, and MIDI editing.

I had a few problems, but I may be one of the lucky ones. I have no real statistics on this, but a large number of Pro Audio users seem to have had very lackluster results from SONAR. Most notably to me are the number of users who were getting a decent track count in Pro Audio 9 who cannot get SONAR to play back their tracks without dropouts, especially considering that my performance and track count actually improved in SONAR.

It’s difficult to know what’s going on. Some of these people are probably running the wrong drivers for their OS. Others haven’t experimented with SONAR’s audio engine to see if adjustments would improve the performance. Still others may be having hardware incompatibilities between new WDM drivers and their existing hardware. Others made the switch to Windows 2000 prematurely and may have other hardware drivers which are not Windows 2000 capable. I have worked with a few of the people who are having the worst problems with the software, and in every case I found issues that should have been obvious – running the wrong drivers, running new (untested) hardware, making changes to the OS, drivers, motherboard, and application simultaneously, and the list goes on.

And there’s the rub. There’s enough new technology under SONAR’ s hood – particularly WDM and DXi – that we are now going to get to work through a new generation of hardware / software compatibility issues. I think that enough users are able to get the kind of performance that I am getting from SONAR to justify the switch, however some people are going to get burned badly by this new technology until the problems shake out. But there’s no reason to get burned badly. You can install and use SONAR with MME drivers today, and all you’ll be missing are live softsynths and live effects processing.

If people work through their compatibility problems, they will probably be rewarded with a better system than they started with. But some people can’t afford to work through hardware and software compatibility issues. To those people, I strongly recommend that they really do their homework before plunging into SONAR, Windows 2000, and WDM, and make sure that everything is going to work right before switching over. To those people, I will share two words of wisdom: dual boot. Later in the article we’ll discuss the advantages of setting up SONAR in a dual boot environment.

Other Shortcomings

Earlier I praised SONAR’s use of envelope automation. However, SONAR’s envelopes suffer tremendously from poor integration with the audio engine: any change to any audio envelope results in a brief stopping of audio, also called a gap. This gapping is very painful when trying to mix a song using SONAR’s envelopes. By contrast, Vegas almost never gaps: if you move an envelope you’ll only get gapping if you change the envelope’s position at the current point in time. Cakewalk has to repair this deficiency promptly or its users will rebel against the envelopes, effectively throwing out the baby with the bathwater.

I really like SONAR’s looping capability. What is interesting is that some loops actually play better under SONAR than Acid, while others are worse under SONAR compared to Acid. What is not interesting is that SONAR lacks the ability to save Acidized audio. If you like to create your own loops, you’ll still need Acid. This is a shortcoming that should be addressed quickly. As an individual that likes to cull through existing audio and harvest loops, I for one would make tremendous use of the ability to Acidize and save loops directly in SONAR. Cakewalk, add this capability.

SONAR, like Pro Audio, makes use of internal file naming conventions when recording audio. So, for example, your bass track isn’t saved as “ Bass Take 2” or even “Track 8 Take 2” but rather “ C6534PW4.WAV”. If you use SONAR like a big multitrack tape machine you may never notice – I’ve done a number of album projects like this and couldn’t care less what the files are named. However, if you remix, do any video post, or need to move your audio around among various applications, SONAR’s file naming scheme may prove to be a hindrance. We would have liked to see a customizable file naming convention in SONAR.

Kudos

The pain that some users have gone through has not gone unnoticed by Cakewalk. After releasing SONAR in April, Cakewalk quickly released two application patches to address the most egregious and easily repairable bugs. And our close experience with Cakewalk suggests that yet another, more comprehensive update is in the works which will address tougher usability problems free of charge to registered SONAR users, though the specifics of the functionality and timing of the interim release are not yet known.

“Yeah,” you comment, “just another case of letting the users find the bugs after they pay for the software.” A valid comment, perhaps. However, a few points should be made. First off, experienced professionals know to be wary of version 1.0 releases of anything. Secondly, SONAR’s support for brand-new driver and DX technology simply guarantees a level of “mystery bugs” that aren’t going to be discovered until enough people are using the software. Finally, in the end, all software has bugs. It’s what happens after the bugs are found that really counts. Cakewalk has announced that it’s going to work through the problems that are on the table. That’s what really counts.

ProRec commends Cakewalk for standing behind its paying customers and continuing to work through and solve problems even months after the software has shipped. Few other companies offer this kind of dedication to their customers. Kudos to Cakewalk for going the extra mile.

Summary

I have to say I’m fairly surprised to come out in favor of SONAR. Obviously the feature set is leading edge. But I suspected that stability and performance would suffer greatly with this tool. I have discovered that with the right combination of compatible hardware and drivers, performance is greatly enhanced, and stability is nearly up to Pro Audio 9 standards. It’s a far cry from the days when users struggled through buggy, half-baked implementations like Pro Audio 7. This is good stuff, and with another round of updates, SONAR will be up to any challenge.

SONAR is a cornucopia of powerful features. If you can make the switch to SONAR and take advantage of what it offers, then you will be both pleased and amazed by its capabilities. At $299 street, or $99 upgrade, the capabilities are much more than worth the cost in time and money required to make the change. Most users, like myself, have been able to make the switch with a minimum of heartache and are already reaping the many benefits of this powerful tool. If you’re like me, you’ll hit the ground running with SONAR and never look back.

If your hardware and software will permit the change to Windows 2000, here is my direct advice: get SONAR now, and set it up in a dual-boot environment in Windows 2000. Setting up a dual-boot environment is as easy as checking a box when you install Windows 2000, and lets you safely retain your existing Windows 98 system and applications. Dual-boot will provide you a way to take your time making the switch to Windows 2000 and will let you retain a safe working platform while making the transition. When you’re up and running in SONAR with Windows 2000 and WDM, you can simply remove the Windows 98 installation. Piece of cake.

The future of audio and MIDI multitracking is here. With a feature set that meets or exceeds tools twice the price, as well as incredible performance and world-class support and customer care, Cakewalk has raised the bar for the entire computer-based recording industry. Now that PC users can run audio applications with confidence on Windows 2000 we can start to reap the benefits of stability and performance it offers. And now that audio applications do not have to rely on proprietary audio driver technology in order to provide optimum latency and performance, we should see a convergence towards WDM eventually resulting in the best possible performance, stability, and reliability for the entire industry.

The technological change embodied in SONAR is a milestone for the audio industry. I predict that a year from now Windows 2000 (or XP) and WDM drivers will be considered the new standard for audio platforms – with SONAR leading the revolution. Cakewalk, once a technological follower, has pushed the technology envelope with SONAR. It is now for other companies to catch up.

Leave a Reply