Well, your decision path is more like this:
Should you buy a sound card and Cubase (or another package like Sonar or Vegas (PC) or Logic or Digital Performer (Mac)
or
should you buy a Digidesign interface (like the 001, the MBox, or the 002) to get Pro Tools. The only version of Pro Tools designed to work with other people's hardware is Pro Tools Free (an 8 channel version) -- which is based on PT version 5.0.1 and limited in the OS's it will work with (Windows 98/ME or Mac OS v8.6 & v9.x Other restrictions apply on the Mac version
http://www.digidesign.com/compato/mainfreemac.html
Obviously, there's a huge amount of hype in the industry on PT -- and a PT/HD (their high end hardware based system) is a very powerful system from many angles.
Users are somewhat mixed on PT software, however. Some love it but many others who have used desktop multitracking software from other vendors are less enthusiastic.
I would definitely recommend an 001 over an Mbox if you go that way. The Mbox is okay for laptop field work but because of USB issues has a serious through-the-box latency issue. The 002 is several thousand dollars and I have heard absolutely no word of mouth one way or the other. The 002 is Firewire and is also subject to latency issues although, as I understand it, your 'zero latency' direct monitoring (pre-converter monitoring like Sound Blasters use) options are greater with the 002 (I
believe you can add effects like reverb to the tracks being recorded to avoid "dry monitoring" -- which you're stuck with with an Mbox.
Because the 001 sits in the PCI bus it is not subject to the long throughput and buffering times associated with USB and FW interfaces. (USB "1" is usually limited in the number of tracks that can be recorded at once... most USB cards have only stereo in but there are several that try to get 4 channels in while still maintaining monitoring capability. Firewire has much greater throughput and can handle oodles of tracks -- but because FW has large buffers that must fill up before the signal can get through there is still a serious latency issue.
Still, with 'direct monitoring' that latency should not be much of an issue (with the exception of computer based monitoring FX) -- unless you plan on working with soft synths. (Both USB and FW "time stamp" the incoming audio signal to allow it to be properly aligned with existing tracks, so latency doesn't apply to
that issue.*)
If you're going to be doing soft synths in real time you must avoid USB or FW converters. There is simply no way for the sound to get through the computer (all soft synths have a certain amount of processing latency but it's usually not too bad on today's fast CPUs) and out to the sound interface via the USB or FW (and
that's where the serious latency comes from) and stay in time with other elements. If you knew what your monitoring latency was going to be you could offset that using MIDI for robotic playback -- but, unfortunately it's very hard to predict the precise latency of USB and FW from moment to moment so that doesn't help much. (I'm not that familiar with softsynth/USB/FW issues so there may be others who can fill us in more.)
One last note: as I mentioned, FW and USB latency affects monitoring latency (forcing you to use pre-converter direct monitoring) but the timestamping will keep the audio aligned...
BUT -- and this is a big one -- Pro Tools' existing software (new version due any time) DOES NOT have a
conversion-latency offset built in and you must manually shift each subsequent overdubbed track by a small amount so that the conversion latency (the actual A/D processing of the Digidesign hardware) does not add up or slur the rhythmic precision of your work.
The 001 has a relatively short latency of 51 samples (about 1.1 ms at 44.1 kHz sample rate) and, depending on how you work and what tracks you use as a rhythmic reference you may not even notice the latency build-up or develop much slur.
The Mbox on the other hand has a conversion latency of 164 samples (if I recall correctly) which is about 3.7 ms. If you were doing a full 24 track project (unlike other multitrack software, PT LE limits you to 24 tracks... to get better you'll have to pop the $10K - $15K for a PT HD system) -- and you took your timing cues from each immediately previoulsy recorded track (rather from, say, a click track on the first track you recorded) your 24th track would be nearly a tenth of a second behind your first! But even if you try to take your cues from a click track, you might become confused by subsequently recorded tracks and your rhythmic center could drift noticeably.
Digidesign initially refused to acknowledge this issue but after much flame and discussion on the Digi User Conference board they issued a set of recommendations (manually move each track 164 samples toward the start time) -- and they also issued a memo that emphasized that MBox users should mute the monitor of the track being recorded in Pro Tools so as to avoid the huge USB monitoring latency (which sounds like an slap back echo) and only use the "blend" control on the MBox hardware (basically a crossfader that balances the sound of whatever is plugged into the input with the sound coming out of the D/A converters. Unfortunately Digi provided no way to mute this input short of manually turning down the "blend" control when not recording. You can read about this and many of the other problems with the Mbox in the
Electronic Musician review.)
Digi apparently has NOT acknowledged the conversion latency issue with regard to the 001 but that has been independently verified by numerous users as 51 samples. I have seen no word on what either the conversion or monitoring latency of the 002 is.
You might be asking yourself, if Digidesign is such a brain trust and has such hot technology, why haven't they built in the ability to automatically adust for conversion latency (as virtually all other digital multrack software vendors have) or even come clean on what the issues are with their hardare. And, frankly, I think that's a
real good question.
[CORRECTION: DOH!!! It appears I was wrong about
latency offset adjustment in my own multitracker, Sonar, and quite possibly about other multi-trackers. I'm waiting for confirmation and further information on this... but from my own testing, I'm seeing a 189 sample -- about 4.3 ms -- roundtrip 'lag' when I copy a track via analog out back into analog in... which appears to mirror the behavior of PT LE. More later.
Mega mea culpa!]