We are Ghost in the Machine, a techno/industrial/modern rock band from Wisconsin (USA). We have released and completely produced three CDs: our debut EP, Ghost in the Machine, a remix CD maxi-single, Everyone is Dead/Do You Believe?, and a full length CD, The Haunting Begins.... Currently, our CD, The Haunting Begins..., is carried by the international distributor Valley Media Inc. and is available at online music retailers like Amazon, CDNOW, etc. We are also writers that contribute to music library publishers, as well as audio commercials and similar A/V projects.

We are faithful surfers of the Independent Music Site and have had all of our CDs reviewed by the staff. We were flattered when Dave M. asked us to contribute an article. It is an honor for us to submit our practices and ideas. So, we are going to share with you our studio setup, instrument configurations and songwriting process.

---

Our studio setup consists of many items. For starters, we have a PC with 440BX motherboard, Pentium III 500 MHz processor, 256 MB of SDRAM, Adaptec 2940 U2W SCSI controller, 14 GB Seagate Cheetah hard drive, two CD writers, 32 MB AGP TNT2 video card, Turtle Beach Fiji sound card, Creamware Pulsar sound system, and a network card to connect to a LAN. The Pulsar sound system is mostly the core of our setup as it has DSP power to run virtual synths, mixer, effects and more. I'll talk more in detail about the Pulsar sound system a little later. For sequencing MIDI and audio, we use Cakewalk Pro Audio v9. For audio editing, we use Sonic Foundry Sound Forge. We have a slew of DirectX plugins that are compatible with both Cakewalk and Sound Forge for audio processing. On occasion, we also use synthesis programs like Vaz and Rubber Duck.

There are also several non-computer pieces of equipment that we use. For MIDI controlled sound generation we have a Yamaha MU80, Emu Orbit V2, and Novation Drumstation. Our MIDI controller is a Fatar Studio Logic SL-161. We have several guitars that use in recording; Gibson Explorer, Fender Stratocaster, and Fender Precision Bass. Additional, there is a Crybaby 535 pedal. An Auto-Com Behringer compressor is used to control the dynamics of all instruments that are recorded in to the computer. An Event Rode NT1 large diaphragm condenser mic in conjunction with an Art tube pre-amp for power are used to record acoustic "sounds". To tie it all together we use a Sampson PL1602 line mixer, Behringer PowerPoint headphone amp, Sony MDR-7506 Pro Headphones, patch bays, and Event 20/20p powered near field monitors.

Our basic studio setup is all the tone generators and computer audio outputs are connected to the line mixer, whose outputs are connected to the near field monitors. This allows us to rough mix all the elements together for writing purposes. The MIDI keyboard controller is plugged into anyone of the Orbit, MU80, Novation, Pulsar, or Fiji MIDI port at any given time. It is used more or less as a trigger to test sounds easily or play melodies/rhythms quickly. To do this with the computer mouse is just too time consuming and uncomfortable. Each of the tone generators is connected to the computer by their own MIDI ports via the Fiji or Pulsar systems.

All recorded instruments, whether it is one of the tone generators, vocals, guitar, bass, or other noises, first go through the compressor before the analog to digital converters in either of the sound boards. The reason for this is that by controlling the dynamics of the recorded sound, we can get a better idea of whether or not a "take" was good or not, especially vocals. We find this more reliable than just recording and then first compressing in the digital domain. Also, this helps reduce the possibility of hitting "digital 0" and over loading the analog to digital converters on the sound cards.

Now we'll take the time to talk a little more about the Creamware Pulsar sound system. First let me say that, no, we are not Creamware employees, but we do absolutely love the Pulsar system and would definitely recommend it to someone who is in a similar studio situation to ours. Our situation was that we had a computer and Cakewalk - we wanted to mix and process audio tracks, and have software synths of sound quality equal to hardware synths - all in real-time and for a moderate price! It was either an all hardware setup (big bucks), a Mac and ProTools (big bucks) or Pulsar. Another attractive feature was that Pulsar is not tied to anyone proprietary sequencer - you can use Cakewalk, Cubase, Logic, etc. ProTools, Paris and some others require you to use their sequencer; not cool, if you have already bought a sequencer.

The Pulsar card is a PCI card with 2 ADAT Lightpipe connecters, MIDI in/out/thru, AES/EBU digital I/O and stereo 20-bit ADC/24-bit DAC analog I/O. The card is capable of 16 channels of 24 bit audio through an I/O converter box via the lightpipe connectors. The main attraction of this device is the fact that it has 4 SHARC DSP chips which provide the processing power for effects, synths and devices. In other words, everything that you hear coming from this card is processed (and synthesized in the case of the Pulsar synths) on the card and not the main CPU. This allows us greater real-time flexibility, in terms of the number and types of effects, synths and devices that we can run, that otherwise we would not be able to do in software only situations. If you run out of DSP power and have extra cash, you can always purchase a DSP expansion board and connect it to Pulsar via its TDM bus. Pulsar has a 32 channel fully automizable, functional digital mixer. All audio from an external sequencer or Pulsar synthesizers, in the Pulsar environment, can be routed through it for processing and mixing. There are many Pulsar synthesizers which have various capabilities like LPF's, HPF's, LFO's, ASDR envelope generators, etc. The sounds of the synths provided by Pulsar are phenomenal! They compare to some of the most popular hardware synths available (in our opinion). Pulsar even has a modular synth! We can make our own modular synthesizers with the many modules provided. The GUI (graphical user interface) for Pulsar is also excellent. It is a drag and drop environment in which all devices have there own interfaces - everything is MIDI controllable including the mixer and synths! Last but not least, Pulsar has an 11 band vocoder and an Akai sample player! For our situation and setup, Pulsar was the complete solution. For info on Pulsar, check the Creamware hompage

From beginning to end, the songs that we write go through many stages (phases). It all begins in Cakewalk where we use its multitrack capability to arrange our MIDI and digital audio tracks. Many times a song will start as simple drum loop MIDI sequence routed through our Novation Drumstation, Yamaha MU-80, Orbit or the drum synth in Pulsar. It may stay in this form....usually only a few measures long for days, months, even years. Typically, the next thing added is the bassline. The bassline sound is pulled from one of our tone generators or Pulsar synths. At this point, the "song" has no real structure. The loop that we have made may become the chorus, verse, bridge......any part. After that, other instruments are added to the small loop. By this time, either we really like what we have and we start adding to it, or it will sit on the hard drive for awhile, perhaps to be included into another song in the future.

As we begin to add on to the song, the different parts of the song, verse, chorus, bridge and breaks begin to take shape. At this point everything is MID with exception of sample-like audio. The song has a basic drum track, a bassline and probably a few other lead synth type instruments. We typically change any patches or synth settings from this point forward. Perhaps we want a thicker, lower bass for the bassline, now that we have an entire song and this type of bass suits it better. Perhaps we decide to switch from a piano sound to a synth for one of the leads. Although it is not always the case, often, many of our sounds are in a state of flux up until we actually record them. At this stage, we have a full-length song with well-defined parts.

Techniques that we sometimes use to alter a MIDI sound are to use MIDI controllers. We most commonly use them to control volume and pan. However, different devices have different MIDI controllable parameters, for example: filter cutoff frequency and resonance, velocity, attack, sustain, delay, release, etc. For our songs, we control the velocity to give a more human-like, amplitude dynamic, the ASDR to control the amplitude envelope of the sound, the filter cutoff frequency to give the "instrument" the appearance of varying sonic depth, etc. Occasionally, we use one MIDI controller to control another MIDI controller. For example, the value of the velocity controller will modulate the filter cutoff frequency. By doing this, the amount the filter opens or closes is directly dependent on the value of the velocity controller.

The next stage is where we add the highlighting instruments. These range from a light warbling in the background to additional percussion that gives the song extra punch in areas that need it. We now have the song at the stage where we can add lyrics to it. In some cases, the lyrics had been written specifically for it, other times we try out lyrics that we have that seem to fit the mood and/or structure of the song. The lyrics then go through a testing phase to see if they need to be changed around at all to fit the song. To test the lyrics we practice them while the MIDI portion of the song plays. After a few weeks of practicing it, we generally have a good idea on what the main vocal melody will be.

Up to this point, everything has been done in MIDI. Depending on what type of song it is, samples and other audio blurbs will be added at this point. We also are adding any finishing touches to the song and are getting a rough mix. The last step before we convert all of the MIDI to digital audio is to record the vocals. We have employed many different methods when recording vocals. One method that we have used is to record a full verse then go back and find the weak spots and then record only the needed portions. This is an effective method if you have very little hard disk space. If there is a lot of space available on the HD, then what we have done is to record the vocals one or two lines at a time saving any good takes, or the first 3-5 that sound decent. We would then go back and compare the different takes and keep the best one, or combine the best of the different takes until we had a full line.

After we have decided what to keep on all vocal tracks, it's time to convert all of the MIDI into digital audio. Before we actually record the MIDI tracks, we separate all distinct MIDI sounds to their own track. This is done mainly with drums. For example, we move the snare to its own track, the bass drum to another track, etc. The reason we do this is to isolate each sound in its own digital audio file so that we have more control over the sound of each drum. If we were to leave the drums as just one digital audio file and wished to boost the bass drum's resonant frequencies, it would have an unwanted effect on the other drums, making the cymbals muddier in this example. In our past work, once everything is separated, we then record each track separately. This is a very time consuming process, so for the next CD we plan to have a multiple I/O box, in conjunction with Pulsar, to speed up the recording process.

Once all of the tracks have been recorded, we are ready to move on to mixing. For us, there are two main aspects of mixing, horizontal positioning in the stereo field, which is pan, and radial distance from the listener, which is stereo depth. Panning is important to get good spatial separation of instruments, so that they can be heard clearly. Depth in the stereo field is achieved by adding reverb. The more reverb on an instrument, the further away it sounds, therefore if you want something to sound "in your face" it should have very little reverb. The visualization we use in mixing is that of a live rock band. This means that the center of our stereo field corresponds with the kick drum. Other drums and instruments are then panned in the stereo field, about its center, as they usually positioned on a stage. The lead vocals are usually placed approximately in the center and up front. The stereo positioning of vocal harmonies is determined by their number, the mood of the song and the location of other instruments of similar frequency responses. Alternatively, some songs do not lend themselves to this type of stereo positioning. Unorthodox techniques of panning and depth are used in these experimental and esoteric cases.

Part of mixing involves effects processing. The effects that we use fall into three categories (or a combination thereof): those that effect the frequency, such as pitch change, and those that effect the time length, such as delay and those that effect the amplitude, such as compression. With any type of effect, we are trying to pronounce, correct, or create special effects. Examples of frequency domain effects: When applying EQ, we boost resonant frequencies and attenuate those frequencies which conflict with other instruments - this gives them frequency separation. When applying pitch correction, we pitch shift those parts of the digital audio which are out of tune. Pitch shift can also be used as a special effect. An example of this would be on a vocal track to make it sound like Satan or the Chipmunks. Time domain effects are used to fatten, contact or expand and also create special effects. Amplitude modulation effects are used to control dynamics and create special effects, for example, ring modulation.

After the mix sounds perfect to our ears, everything is mixed down to one stereo track. Once the song is in this form, we can pre-master it. Really the only steps involved are to compress the track to even out the loud and soft parts and add percussive punch to the mix. If the song is very dynamic, moving from loud to soft quite often, and it is designed to do this, then we wouldn't want to compress it too much, or we would lose these volume changes. Once the song is compressed, we EQ it to give the high and low end a boost and the low mids attenuation. Generally, we try to have all of the songs sound similar in terms of the frequency envelope or response. The amount of EQ applied to a given frequency bandwidth in each song depends on how different it is from the other songs. The last stage is to limit and maximize the amplitude of the song so that it is as close to digital zero or 0dB as possible. We usually try to match it up against professionally produced CDs. After the song is limited, it's ready to go !!

More Information about Ghost in the Machine is available at :
http://www.gitm.net