My "studio" is based round my PC and a set of external modules and other devices in a trolly rack. Basically, the PC is used for control (sequencing) and production (recording). The rack is used for sound generation, MIDI interface, digital signal handling and mixing. Put another way, the PC generates MIDI and sends to the rack and the rack converts to digital audio which is mixed in the MOTU mixer in the rack. The PC also uses software synthesisers which send audio to the MOTU mixer. The output of the mixer is sent to a an external DAC for monitoring and back to the PC for recording. Some of the equipment is new and some second hand. Some of it is the remnants of older setups, so the organisation doesn't necessarily make a lot of sense. I wouldn't do it this way if I were building from scratch.
The rack is a QuikLok trolly rack. It has 10U spaces and castors. The idea was that I would be able to move it around and this would make it easier to get at the back when I needed to. Actually, it is rarely moved these days, partly due to the mess of cables at the back. I could also add additional parts to bring it up to 30U, but don't have the need (or the living space) at present. I have, however, extended by 3U by the simple expediant of attaching some rack rail.
The aim of the rack was to keep as much of the sound production outside the PC as possible. As you can see below, I have used a variety of sound cards in the PC from time to time, but I always had major problems keeping the noise level down. These days I work entirely in the digital domain and so those problems have largely gone away. The noise problem is now chiefly acoustic :-(
PC noise is a major bane of my life. My first studio PC was a dual Xeon box with Yamaha DSP factory devices. One of the main audio cards in that PC was a Yamaha AX16-AT. There are a ton of heat problems with this card. It only operates up to 40C. This is a normal PC operating temperature. So I had to install extra fans in the PC. I then had a constant battle with fan noise. I installed ultra-quiet fans and an external fan controller. The controller allowed me to manually speed up and slow down the fans to optimise the noise and had very distracting bright blue lights on the front of the box. But, it was still a hot box. It was a relief to get the headphones on sometimes.
So, after a while I rebuilt my studio PC focusing on making it quieter. I have always built my own PCs and this was the latest in a long line of boxes. The new PC was a 2.0GHz Core Duo on an AOpen MicroATX motherboard (i945GTm-VHL) with built-in Dual SATA/LAN/Graphics, etc. and a SATA hard disk (250GB). It was all located in a modded Antec Aria chassis. The main modification is that is used an external DC power supply reducing the heat it generates. As a box, it was OK for a while. There were things I liked and things I didn't like. I liked locating everything on the rack. So, on my desktop I just have my monitor (a 20" Dell 2001FP flat screen), keyboard, mouse and, sometimes, my Oxygen 49 keyboard. For backup, I had a DVD-RW drive that can back up 8.7GB on dual layer disks. Everything else was on the rack. Despite the new PC being quieter, I never really felt it integrated in properly, and so after three or four years I decided to do another rebuild. Bear in mind, this is a hobby for me. Anyway, the pattern of design and implementation was much the same and I will probably write it all up in detail someday.
The current PC is based around an XCase short form rack installed directly into my trolley rack. The motherboard is an Asus P7-H55D-PRO with an Intel i5-660 dual core with hyperthreading. The main disks are a 250GB SSD system disk. I have a RocketRAID RAID controller with an IcyDock MB994SP-4S housing a separate 1TB of disk used for data and archives (RAID 0/1). I have a Blu-Ray dual layer reader/writer drive allowing to backup up to 25GB. When I build a new PC, I tend to use lot of parts from the previous one and so my system is a bit of a mish-mash. So, inside the box is the the software synthesisers and Sonar 8.5 sequencer. As before, I have my monitors, keyboard, mouse, etc. My Oxygen 49 keyboard is located on a laptop arm which I can swing in and out of the way as necessary. Generally, I write my music by hand, but sometimes it's handy to have a keyboard.
The PC is connected to the "house LAN" via a firewall and from there to the Internet using broadband. My new PC sits on the rack, which is handy because it massively reduces the number of cables to my desktop. The PC tends to be the hub of everything I do.
I run Win7 (32-bit) using the ACPI MULTIPROC HAL. The PC runs Sonar 8.5 Professional for sequencing. I tried Sonar X1C and didn't get on with it and reverted. The big monitor really helps here. I mainly use the MIDI features of this software. As far as DXi plug-ins are concerned, the only ones I run are Native Instruments Kontakt 2 as a soft sampler, Atmosphere and Omnisphere for pads, Trillian for basses. I am using more and more Software Synthesis and now the battle is to have more and more speed. I suspect in the future, we will be doing everything in the PC, but I have listened to lots of soft synths and compared them to external modules and the latter nearly always wins hands down, though there are exceptions to this. It might simply be that the basic sounds available on the external modules is, simply, a higher quality. However, I also think trying to make a general purpose PC processor behave like a DSP (or, in some cases, custom ASICs) is pushing the edge of current technology. And as general purpose processors get faster, so do DSPs and that alone has allowed them to maintain an edge. In the end, though, general purpose processors will get fast enough.
I have used Sonar and, previously, Cakewalk for years and have found it to be really reliable for what I do. Also, I have all the instrument definitions done now which makes it really easy to use. I have a number of gripes with Sonar some small, others not. I like to have blank tracks in the track view to allow me to organise the various parts. It's really helpful if you have a 24-part score. This used to be possible in Cakewalk, but was removed in Sonar. I now have to use an empty audio track with -INF volume. There are some stupid things like setting to track name to "Track n" by default. They should use the patch name, if the track name is unassigned. Much more sensible. I always have one track per MIDI channel. If I view the event list, the channel for each event is often different from the channel assigned to the track. I have never worked out what the effect of this is, but it seems wrong to me. I frequently find myself using the "Interpolate" tool to reset the channel numbers (must write a CAL script some time!). What I would really like them to do is implement a way to be able to restrict the channel list menu, so it only shows me what I haven't used. Integrating VB would be really neat as well. I have done this using VBS and VBA in different projects and it isn't that hard to do. Also, the amount of time it takes bringing up the staff editor is getting longer with every release. I must look into this some time, its probably font-related and specific to my setup.
In hardware terms, Sonar is talking to the rack via a MIDISport 8x8 interface. I upgraded a MIDISport 4x4 to this, because I found the 4x4 to be as reliable as anything.
My rack modules are a Yamaha MU100R, a Roland XV-5080 and a couple of Emu modules.
The MU100R gives me 128 voices on 31 channels of AWM-based samples with a powerful effects unit built-in. I tend to mainly use the "LCR Delay" effect (as well as reverb and chorus) on one or, maybe, two sounds in the mix and control the Wet/Dry levels directly from Sonar. I don't use any of its audio features though. Generally, I like this synth for a few of the pads, the synth basses and the kits.
The MU100R has a PLG-150DX daughterboard installed. This is basically a DX7 on a card. It is really handy for hits and "special" sounds. Most of my pieces have something coming from there. It is nice to have a non-sample based synthesiser in the arsenal just for variety. Like all FM synths, it is a nightmare to program, though. In common with all my modules, special stuff is set up on channel 16. In the case of the Yamaha, this is where the DX is. I take the audio output from the MU100R and DX via an ADC to give me a digital feed.
The 5080 is largely used for pads, strings, world sounds and so on. Brilliant! It is expanded with the Orchestral (SRX-06), World (SRX-09), Piano (SRX-11) and Complete Orchestra (SRX-06) collections, so some room for expansion, when I get any money. I tend to use a single performance I have created that maps all MIDI ports to their respective channels and has MFX A, B and C dedicated to parts 14, 15 and 16 of port B respectively. This allows me to control the use of MFX for particular parts from Sonar. Most sounds I use with no MFX at all, because I find that too much FX muddies the mix. My gripes with this module include the annoying habit of the presets to wander around the stereo field, the lack of time control in the SRX expansion "menu" presets, an irritating DC offset in the output and the boot time. Also, it has a habit of dominating the mix in a way the Emu doesn't, especially where FX is being used. I tend to keep the volume very low and heavily layer the parts. This module is something I would definitely buy again.
I have a growing number of patches using my own samples. I have also installed a SmartCard and have downloaded and registered various preset collections from Lilchips. Very handy. Thanks, David.
The Emu boxes are a second-hand P2K and and Orbit 3 module. The P2K is expanded with a Protozoa ROM, Siedlacek and World ROMs. There is a lot of useful stuff on the Protozoa and World ROMs and the Protozoa is so much cheaper than the others. I am less impressed with the Composer and Siedlacek ROMs though. I tend to use those less than the Protozoa. The Orbit 3 is the most recent addition. People tend to either love this thing or hate it. I love it. Rob Papen's combination of electronic and acoutic sounds is really special. I don't really use Superbeats though. I prefer keeping this sort of stuff in Sonar. It is expanded with the Vintage and X-Lead ROMs. If I were buying again, I'd still go for the Orbit, but I would probably get a V2K rather than a P2K and add the Protozoa.
I love the way that Emu stuff fits into mixes really well. However, I really hate the way the presets are set up for FX. You use a patch with FX on one part and it mucks up all the others. I would set up a sysex to set all other parts to dry, but sometimes you want it and sometimes you don't, if you see what I mean. Consequently, I have a single "multi" set up for both modules that uses the FX for the patch selected on channel 16 (again). In Sonar, I set up a specific track for channel 16 dedicated to selecting the FX, but with no sounds on it, just to remind me of the effect of using it. I have also setup the Emu boxes to output all channels to Send4, (most patches bypass the FX unit on this channel). Channels 8-16 on port A, which are setup to output to Send1 (via FX). If I want to use FX, I select the appropriate channel in Sonar. It's not completely satisfactory though. I wish Emu had separated the special FX from reverb and chorus and have patch defaults that use the former and set the send level to the latter to zero like Yamaha do. Lot's of the patches on the Emu module need a little reverb for warmth. Then, when I want FX on a patch, I can just turn the send level up on the appropriate parts using a controller.
I have had Kontakt 2 for a while mainly because I wanted to "get into" sampling. As I said before, I mainly use it for "special" sounds. I have sample libraries from Spectrasonics and Roland that I have converted (and adapted) to use with the Kontakt. The main library I use is "Symphony of Voices". It is full of vocal sounds that sound really brilliant. I also have "Vocal Planet" but I use that less.
I purchased Atmosphere a few years ago for pads. I love the really complex motion pads, though sometimes it's hard to integrate into my tracks. It's a work in progress. I have, very recently, upgraded it to Omnisphere. It's a wonderful thing, but I have a suspicion that it may suffer from the same problems. I also have Trillian which is essentially the same synth engine as Omnisphere, but geared to all kinds of bass. It's awesome.
Part of the problem with DXi and VST is that they offer a poor degree of integration with the host sequencer. For instance, they do not "export" patch lists, controllers, etc. into the sequencer environment forcing you to use custom interfaces for each "device". It would be so much easier if you could do it directly from Sonar when you are just in the "pick and choose" mode.
I use the XV-Editor to play with the sounds on the Roland. But, only rarely, as the start-up times are so long. I also use XG-Edit & DX-Simulator for the Yamaha cards. For the Emu modules, I wrote a patch editor called Proteum. You can find out more about this here. What I would really like to have is plug-ins for Sonar that do this job for each module. I know it can be done, but Emu and Roland simply don't put in the effort. They don't even supply Instrument Definitions, let alone Sonar plug-ins. Maybe there isn't any money in it.
I take direct digital outputs of the rack synths and send them to the RME ADI-8 DD unit using SPDIF (24/44.1). The RME unit re-clocks the data againsts its internal reference clock and converts it to ADAT format. This is sent via lightpipe to the mixer. The RME box was expensive, but the clocking is so solid I dont have to think about digital clocks at all and can treat the digital feeds just like you would analog ones. Unfortunately, it only does 4 inputs so I have run out now. The RME also provides the reference clock for the mixer.
The RME box is there really for historical reasons and though it is neat, you wouldn't do it this way if you were starting from scratch. However, I got it because I needed SPDIF to ADAT conversion as well as the reclocking function.
I tend to run the whole system at 24-bit 44.1kHz. I use 44.1kHz because my final output is CD and I don't want to introduce more reclocking later on. I use 24-bit because that is the maximum rate supported by all my equipment. I think the word width is more important that the bit rate, so given a choice between going from 96kHz/32-bit to 192kHz/32-bit or 96kHz/64-bit I would opt for the latter. I think you can hear the difference during fade outs and as reverb dies away. Anyway, this is not an option for me.
The output of the RME box is sent to an MOTU 828x rack-based mixer. As I said earlier, I used to use a set of Yamaha DSP Factory cards, but I found, given my limited use of the DSP Factory, that I could replace them all with a Emu card (1212M). However, as time progressed, I foudnthe card was making it hard to upgrade the PC. So I got rid of the 1212M and replaced with the MOTU mixer. The mixer mixes the ADAT inputs with PC sound sources and outputs the mix to files and the SPDIF output. Because the mixer is an external device, it consumes very little CPU from the PC.
During everyday use, I monitor the sound through headphones. These are plugged into an external Extigy box that takes the SPDIF feed from the mixer. I used to use Technics headphones, but the clarity left a lot to be desired. I then bought some Sennheiser HD-500 that were much better though the frequency response curve is not very flat, particularly at the bass end of the spectrum. Also, they are open-backed and, therefore, its very noisy for other people in the room. I purchased some Sony MDRV-7509s. These were brilliantly clear but they died after a couple of years. I am currently using a pair of DT-770 Pro headphones and these are OK, but you need to keep moving the headband because they can makethe top of my head sore after a few hours.
When I come to do a production, I mix down to 24-bit 44.1kHz standard PC wave files using Audacity. I also use Audacity for post-production is done on the PC. Actually, mostly, I just deal with the starts and ends of tracks, a little normalisation and so on. Then I burn directly to CD using Nero.
There is a cable box full of... you have it... cables. It also has things like bits from the end of lightpipes, some rack fittings and so on. All the usual stuff you can't find when you need it. There are also some very cheap old digital cables. This is a big waste on money, never use cheap digital cables, 'cos you'll only have to spend the money again on decent ones. Mind you, I am not advocating spending hundreds of pounds on each cable as some do. I am not convinced it makes this much difference, at least, I can't hear it. I think a middle way of moderately priced ones is the way to go.
In the old days, I worked wholly with PC sound cards. I got through quite a few of them in my time. My first was a WaveBlaster daughterboard hanging off a SoundBlaster card. If you didn't know, the WaveBlasters along with a lot of Creative stuff was designed by Emu. It was actually a Proteus 1/XR on a card. Very handy. Later, I added a Turtle Beach Maui card, adding the ability to do sampling, though I never really used it that much. I always had difficulties with the drivers for the TB stuff. At other times, I have had Terratec DMXFire (weak), TB Multisound and Santa Cruz (more driver problems) and a Yamaha DB-50XG cards (very handy). The Yamaha card was one of my favourites. And, as I said, there was the DSP factory as well.
The earlest pieces on this site (Echoes) use the WaveBlaster, the DB50XG and, occasionally, the Maui card. The second CD (Sound Foundry) dispenses with the Maui, but adds the SW1000XG. Bamboo House is almost entirely the SW1000 and the others are pretty much the current setup with the MU100R replacing the SW100XG.
From time to time, I look at remixing the old stuff to suit the current setup, but the results are always less effective that the originals. Maybe, someday...