Mile
Zero

One of the features that Vista was supposed to introduce, and which audio professionals were cautiously optimistic about, was a new kind of driver model meant to provide low-latency, sample-accurate sound. Windows has always had low-latency sound, but it's gone either through the WDM/Kernel-Streaming drivers, which are kind of a hack, or ASIO, which is a third-party driver standard created by DAW maker Steinberg. Windows has never had a built-in, professional-level audio subsystem the way that OS X has Core Audio and Linux has... well, whatever the hot new Linux audio API is this week, most likely JACK on top of ALSA. So an upgraded sound layer for Windows, one that would offer both ubiquity and performance, was welcome news.

Still would be, actually.

Now, I'm generally bullish on Vista, so don't get me wrong: I do appreciate the new sound mixer, which offers per-app mixing and better quality than the old mixer. I dig the multimedia scheduling process, so media playback is more reliable than it was previously. And driver support for the old standards has reached entirely satisfactory levels--I'm not lacking for external interface support any more. But the new driver model, WaveRT, shows no signs of being able to replace ASIO or WDM/KS on my laptop, for a couple of reasons.

The first cause for concern is that the driver model for WaveRT wasn't designed correctly in its initial incarnation. When digital audio is output from a sound card, it comes from the front of a buffer of samples on the card. Applications periodically refill the buffer by adding new samples onto the end. If they don't add samples before the buffer empties, the soundcard runs out of gas, so to speak, and the result is skipping or popping audio playback. Most driver models, like ASIO, include a mechanism for the hardware itself to notify the software when the buffer is running low, so it can be topped off.

WaveRT's original implementation, as these posts from Cakewalk tech guru Neil Borthwick explain, did not use a notification mode for some incomprehensible reason. Instead, it required the audio software to poll the audio hardware--to manually check on the status of the buffer. What sounds pro-active is actually problematic, since real-time audio works typically uses buffers as short as 128 samples, which on CD-quality audio means that the buffer empties itself more than 300 times a second. Under the polling mode, therefore, the CPU spends a large amount of time checking the status of the sample supply instead of using that time to render audio. Inevitably, samples go missing and quality suffers.

Late in Vista's development it became obvious that this polling-based model was unworkable, and Microsoft added a notification mode similar to ASIO or Core Audio. And apparently it works great--although WaveRT is unavailable for FireWire or USB devices, users report that for onboard or PCI-based solutions, they get very stable, very low latencies. Onboard sound, of course, is hardly a substitute for a pro interface, but for mixing/composing on the go, it does just fine. Cheers all around.

Unless, like me, the OEM who built your computer hasn't updated their sound drivers to include notification mode, in which case you're basically back to using WDM/KS and ASIO, if possible. Worst case scenario, and I'm looking right at you, Analog Devices SoundMAX HD, the driver doesn't even support sample rates other than the oddball 48KHz using any driver modes, and so you're basically locked out of using onboard sound for any production at all.

Unlike graphics cards, where drivers are available directly from the hardware manufacturer (with the caveat that a modified INF may be required), audio drivers are usually "firewalled" behind the company that designed the computer. I can get a generic driver from Nvidia for my Quadro 140M card (and I have, since it doubles the framerate in Stalker and Sins over Lenovo's latest version), but for audio features I'm basically stuck until Lenovo gets around to it. There is no generic SoundMAX driver to download, a symptom of both the sad state of laptop sound and the lack of a decent audio hacking community.

To their credit, when I filed a request to find out about the availability of updated drivers, Lenovo promptly called me back for clarification and told me that they'd be sending a note on to their technical hotline. It's not as helpful as actually getting my hardware to function properly, but they were at least very nice about it. Score one for the service contract, perhaps.

The blame here is threefold. It's Microsoft's fault that they didn't get WaveRT into workable shape from the start, since audio buffers are a solved problem. It's Analog Device's fault that they either haven't written a compliant driver or they refuse to release it. And it's Lenovo's fault for lagging in their own driver updates. No doubt the audience for Thinkpads is a bit thin among media professionals, making it a low priority for support staff. But the machine is exemplary by other standards: fast, stable, tough as nails. It is unfortunate--although par for the course in this industry--that its audio capabilities can't live up to what's been promised.

With relatively little fanfare, Cakewalk released Sonar 7 this month. It looks like it has some really nice features. I've got the Studio edition, and I'm going to review it for Ars. There's just one problem:

Where to start?

It's possible, with Pro Tools, to feel like it's a very simple program. That's mostly due to the window layout, which is incredibly friendly and simple. Pro Tools rarely (if ever) has more than 3 windows onscreen at any time--Edit, Mix, and Plugin. There are menus and subsections for those windows, but when they added right-click support in 7.3 it meant that you could basically ignore 90% of those. So even though the software is actually fantastically powerful, you don't really notice most of it.

Sonar follows a different kind of metaphor. It's more like Cubase. There are lots and lots of windows and palettes and menus. It doesn't make it hard to use, necessarily. In fact, because almost everything in Sonar can be customized, including the menus, you could probably make it look a lot like Pro Tools. But out of the box, Sonar looks like something complicated, and a little overwhelming. It doesn't let you forget just how much there is.

Because let's face it, these packages--any of them, from Live to Logic--are really unbelievably full-featured. Any tool at this level is. That's one reason why people complain about Word: it is stuffed full of functionality for working with text--so much so that you sometimes can't find the feature you need. Final Cut Pro has the same kind of in-your-face complexity. Don't ever press a button on the keyboard unless you absolutely know what it does in Final Cut. They're all mapped to something. I accidentally hit the keys with my elbow the last time I did an editing project, and it took me five minutes to put everything back.

So how do you review something that big? Where do you start? Surely, I can't review everything in the package. It would take me months. And that's assuming I even know what to do with those features--like Word, each of them is probably used, but perhaps only for certain kinds of projects.

My plan is to use Sonar like I would do my work projects--meaning that I'll do a little radio recording, a little instrument recording, and some midi composition. I'm also going to download some of the NIN multitrack files and load them in as a way to test the mix engine in 32- and 64-bit modes--I doubt any of the material I have lying around is high-fidelity enough to make a difference.

There's also some standard bullet-points that are different from the other DAWs, and I think it's important to test those. So I'll be putting AudioSnap, ACT mapping, sidechaining, and the various VST instruments under as much scrutiny as I can apply. I'm open to other suggestions, but I think those are the main points.

The bad news is that no matter what I do, I'll feel like I haven't done it justice. But the good news is that DAWs have reached the point where the basic functionality is the same from system to system, so I hope most users will be able to fill in the gaps.

How many "pro" musicians are using Garageband as a serious music production tool? Peter at CDM sorts them into a few basic categories in a post on improvements in Garageband '08, which finally includes some features that I would say a real DAW needs to have: plugin automation, multiple time signatures, and multi-take recording. It's still a toy, but it's becoming a decent entry-level recording tool, instead of just another ACID clone.

It's been said that Microsoft should think about offering a comparable production program for Windows, so PC users can also have a (sort-of) free music creation tool. But I'm not so sure that we need it. I'm not terribly conversant with the state of OS X freeware, but there's lots of free software on the PC that offers the same functionality (or more) as Garageband, and after three years of experimentation on the cheap, I think I've probably used most of it. Here's a few good options for newcomers to audio production on Windows:

• Cubase LE - No, it's not technically free, since it only comes with OEM hardware. But look at it this way: on any computer platform, if you're doing more than just dragging loops onto a timeline (and please: we have too much of that already), you're going to need to buy an external interface anyway, because built-in sound cards are really bad input devices. There's a lot of great hardware that comes with Cubase LE, including stuff from Tascam, Lexicon, and Presonus. And this is, in my opinion, as close as you can get to pro software at this price level. It's got full automation, lots of tracks, records up to 8 inputs at a time, uses regular VST plugins, and includes a decent set of built-in effects. Honestly, there's not much missing here.
• Krystal - Krystal is free from Kreatives, and has a lot to recommend it. It's a multitrack recording program with an attractive and straightforward interface and VST support (with four plugins included). It's roughly comparable to Garageband 2, but doesn't have the looping support. It also lacks MIDI functionality, plugin automation, and more complicated routing options, and the input routing is a little weird. But for all that, again, you can do pretty well with Krystal if your goal is simply to record and mix audio. I wouldn't hesitate to use it if I needed to track a band at the last minute, especially coupled with a few tools (freeware drum machines, for example) to supplement its functionality. There's supposed to be an updated version (K2) coming soon, but they've been saying that for a couple of years now.
• Reaper - Like Cubase LE, Reaper isn't technically free. It's supposed to be $40 for personal use, and $200 for commercial purposes. But you can download the shareware version, which has full functionality and never expires--Justin Frankel, the head programmer, is the guy behind WinAmp and Gnutella, and he's not a real stickler about licensing. The good news is that Reaper does everything that Sonar, Cubase, and Pro Tools do, and some things that they don't. Reaper has unbelievably flexible routing options and prides itself on its ability to change almost anything while recording. It shows up as an ASIO sound card for recording from other applications, or it supports ReWire if you want. It's gotten rave reviews from music mags. The installer is only two megs. The only bad news, as far as I'm concerned, is that some parts of Reaper still look and feel a little clumsy. It's skinnable, but that doesn't necessarily make it easier to use. Still, usability has improved immeasurably since I first used it (version .62, I think), so it'll probably only get better. This is an insanely powerful tool, and there's nothing on Mac or PC that competes with it in terms of value.
• Audacity - I include Audacity as a word of warning only. I actually don't recommend using it for recording projects, for a variety of reasons. It doesn't have much in the way of plugin support, it only records in stereo, there's no support for low-latency recording like ASIO, and it distorts at the drop of a hat (I'm guessing it actually mixes at the same bit depth as the project). Audacity is one of those things that people recommend all the time, but really shouldn't. It teaches a lot of bad habits, and there are much better options out there for free.
• Ableton Live Lite - Ableton Live (the full version) is a real paradigm shifter for audio software. Live Lite, the OEM version that comes with just about everything M-Audio makes, lets you see all the great stuff from the full version, but locks a lot of it away from actual use. This is frustrating. The good news is that it's practically free, much like Cubase LE. Buy a MIDI keyboard, and you'll probably get Live Lite. Buy any of Line 6's computer products, and you'll get Live Lite. I've owned four or five of these CDs, at least, and had to give them away to other people--you could probably ask around and get a copy, if you have musician friends. For free, Live Lite's not a bad deal--it's basically a 23rd century four-track tape machine. Ableton just released Live LE to try to cater to the beginner/low-cost market, but it's crippled down to practically the same level. If you're curious about Live, start with Lite and upgrade, instead of LE.

There's also a whole world of standalone, budget (~$100) software "for beginners" out there, but I tend to think that it's kind of a waste, depending on the package. Midrange copies of Sonar or Pro Tools are just not that much more money than something like Sequel or Project 5, and the capabilities they offer go much farther. Even the low-end versions, like Sonar Home Studio, can be pretty good, and the upgrade path is easier on the wallet.