In this section, we’ll look at how MIDI is handled in practice.

First, let’s consider a very simple scenario where you’re generating electronic music in a live performance. In this situation, you need only a MIDI controller and a synthesizer. The controller collects the performance information from the musician and transmits that data to the synthesizer. The synthesizer in turn generates a sound based on the incoming control data. This all happens in real-time, the assumption being that there is no need to record the performance.

Now suppose you want also want to capture the musician’s performance. In this situation, you have two options. The first option involves setting up a microphone and making an audio recording of the sounds produced by the synthesizer during the performance. This option is fine assuming you don’t ever need to change the performance, and you have the resources to deal with the large file size of the digital audio recording.

The second option is simply to capture the MIDI performance data coming from the controller. The advantage here is that the MIDI control messages constitute much less data than the data that would be generated if a synthesizer were to transform the performance into digital audio. Another advantage to storing in MIDI format is that you can go back later and easily change the MIDI messages, which generally is a much easier process than digital audio processing. If the musician played a wrong note, all you need to do is change the data byte representing that note number, and when the stored MIDI control data is played back into the synthesizer, the synthesizer generates the correct sound. In contrast, there’s no easy way to change individual notes in a digital audio recording. Pitch correction plug-ins can be applied to digital audio, but they potentially distort your sound, and sometimes can’t fix the error at all.

So let’s say you go with option two. For this, you need a MIDI sequencer between the controller and the synthesizer. The sequencer captures the MIDI data from the controller and sends it on to the synthesizer. This MIDI data is stored in the computer. Later, the sequencer can recall the stored MIDI data and send it again to the synthesizer, thereby perfectly recreating the original performance.

The next questions to consider are these: Which parts of this setup are hardware and which are software? And how do these components communicate with each other? Four different configurations for linking controllers, sequencers, and synthesizers are diagrammed in Figure 6.28. We’ll describe each of these in turn.

In the early days of MIDI, hardware synthesizers were the norm, and dedicated hardware sequencers also existed, like the one shown in Figure 6.29. Thus, an entire MIDI setup could be accomplished through hardware, as diagrammed in Option 1 of Figure 6.28.

Now that personal computers have ample memory and large external drives, software solutions are more common.   A standard setup is to have a MIDI controller keyboard connected to your computer via USB or through a USB MIDI interface like the one shown in Figure 6.30. Software on the computer serves the role of sequencer and synthesizer. Sometimes one program can serve both roles, as diagrammed in Option 2 of Figure 6.28. This is the case, for example, with both Cakewalk Sonar and Apple Logic, which provide a sequencer and built-in soft synths. Sonar’s sample-based soft synth is called the TTS, shown in Figure 6.31. Because samplers and synthesizers are often made by third party companies and then incorporated into software sequencers, they can be referred to as plug-ins. Logic and Sonar have numerous plug-ins that are automatically installed – for example, the EXS24 sampler (Figure 6.32) and the EFM1 FM synthesizer (Figure 6.33).

[aside]As you work with MIDI and digital audio, you’ll develop a large vocabulary of abbreviations and acronyms. In the area of plug-ins, the abbreviations relate to standardized formats that allow various software components to communicate with each other. VSTi stands for virtual studio technology instrument, created and licensed by Steinberg. This is one of the most widely used formats. Dxi is a plug-in format based on Microsoft Direct X, and is a Windows-based format. AU, standing for audio unit, is a Mac-based format. MAS refers to plug-ins that work with Digital Performer, an audio/MIDI processing system created by the MOTU company. RTAS (Real-Time AudioSuite) is the protocol developed by Digidesin for Pro Tools. You need to know which formats are compatible on which platforms.     You can find the most recent information through the documentation of your software or through on-line sources.[/aside]

Some third-party vendor samplers and synthesizers are not automatically installed with a software sequencer, but they can be added by means of a software wrapper. The software wrapper makes it possible for the plug-in to run natively inside the sequencer software. This way you can use the sequencer’s native audio and MIDI engine and avoid the problem of having several programs running at once and having to save your work in multiple formats. Typically what happens is a developer creates a standalone soft synth like the one shown in Figure 6.34. He can then create an Audio Unit wrapper that allows his program to be inserted as an Audio Unit instrument, as shown for Logic in Figure 6.35. He can also create a VSTi wrapper for his synthesizer that allows the program to be inserted as a VSTi instrument in a program like Cakewalk, an MAS wrapper for MOTU Digital Performer, and so forth. A setup like this is shown in Option 3 of Figure 6.28.

An alternative to built-in synths or installed plug-ins is to have more than one program running on your computer, each serving a different function to create the music. An example of such a configuration would be to use Sonar or Logic as your sequencer, and then use Reason to provide a wide array of samplers and synthesizers. This setup introduces a new question: How do the different software programs communicate with each other?

One strategy is to create little software objects that pretend to be MIDI or audio inputs and outputs on the computer. Instead of linking directly to input and output hardware on the computer, you use these software objects as virtual cables. That is, the output from the MIDI sequencer program goes to the input of the software object, and the output of the software object goes to the input of the MIDI synthesis program. The software object functions as a virtual wire between the sequencer and synthesizer. The audio signal output by the soft synth can be routed directly to a physical audio output on your hardware audio interface, to a separate audio recording program, or back into the sequencer program to be stored as sampled audio data. This configuration is diagrammed in Option 4 of Figure 6.28.

An example of this virtual wire strategy is the Rewire technology developed by Propellerhead and used with its sampler/synthesizer program, Reason. Figure 6.36 shows how a track in Sonar can be rewired to connect to a sampler in Reason. The track labeled “MIDI to Reason” has the MIDI controller as its input and Reason as its output. The NN-XT sampler in Reason translates the MIDI commands into digital audio and sends the audio back to the track labeled “Reason to audio.” This track sends the audio output to the sound card.

Other virtual wiring technologies are available. Soundflower is another program for Mac OS X developed by Cycling ’74 that creates virtual audio wires that can be routed between programs. CoreMIDI Virtual Ports are integrated into Apple’s CoreMIDI framework on Mac OS X. A similar technology called MIDI Yoke (developed by a third party) works in the Windows operating systems. Jack is an open source tool that runs on Windows, Mac OS X, and various UNIX platforms to create virtual MIDI and audio objects.

[wpfilebase tag=file id=25 tpl=supplement /]

[wpfilebase tag=file id=43 tpl=supplement /]

Section 6.1.8 covered the various components of a synthesizer. Now that you’ve read about the common objects and parameters available on a synthesizer you should have an idea of what can be done with one. So how do you know which knobs to turn and when? There’s not an easy answer to that question. The thing to remember is that there are no rules. Use your imagination and don’t be afraid to experiment. In time, you’ll develop an instinct for programming the sounds you can hear in your head. Even if you don’t feel like you can create a new sound from scratch, you can easily modify existing patches to your liking, learning to use the controls along the way. For example, if you load up a synthesizer patch and you think the notes cut off too quickly, just increase the release value on the amplitude envelope until it sounds right.

Most synthesizers use obscure values for the various controls, in the sense that it isn’t easy to relate numerical settings to real-world units or phenomena. Reason uses control values from 0 to 127. While this nicely relates to MIDI data values, it doesn’t tell you much about the actual parameter. For example, how long is an attack time of 87? The answer is, it doesn’t really matter. What matters is what it sounds like. Does an attack time of 87 sound too short or too long? While it’s useful to understand what the controller affects, don’t get too caught up in trying to figure out what exact value you’re dialing in when you adjust a certain parameter. Just listen to the sound that comes out of the synthesizer. If it sounds good, it doesn’t matter what number is hiding under the surface. Just remember to save the settings so you don’t lose them.

[separator top=”0″ bottom=”1″ style=”none”]

[wpfilebase tag=file id=23 tpl=supplement /]

Sometimes you may find that you want a certain sound that isn’t available in your sampler. In that case, you may want to create your own sample

If you want to create a sampler patch that sounds like a real instrument, the first thing to do is find someone who has the instrument you’re interested in and get them to play different notes one at a time while you record them. To make sure you don’t have to stretch the pitch too far for any one sample, make sure you get a recording for at least three notes per octave within the instrument’s range.

[wpfilebase tag=file id=136 tpl=supplement /]

Keep in mind that the more samples you have, the more RAM space the sampler requires. If you have 500 MB worth of recorded samples and you want to use them all, the sampler is going to use up 500 MB of RAM on your computer. The trick is finding the right balance between having enough samples so that none of them get stretched unnaturally, but not so many that you use up all the RAM in your computer. As long as you have a real person and a real instrument to record, go ahead and get as many samples as you can. It’s much easier to delete the ones you don’t need than to schedule another recording session to get the two notes you forgot to record.Some instruments can sound different depending on how they are played. For example, a trumpet sounds very different with a mute inserted on the horn. If you want your sampler to be able to create the muted sound, you might be able to mimic it using filters in the sampler, but you’ll get better results by just recording the real trumpet with the mute inserted. Then you can program the sampler to play the muted samples instead of the unmuted ones when it receives a certain MIDI command.

Once you have all your samples recorded, you need to edit them and add all the metadata required by the sampler. In order for the sampler to do what it needs to do with the audio files, the files need to be in an uncompressed file format. Usually this is WAV or AIF format. Some samplers have a limit on the sampling rate they can work with. Make sure you convert the samples to the rate required by the sampler before you try to use them.

[wpfilebase tag=file id=137 tpl=supplement /]

Figure 6.37 shows a loop defined in a sample editing program. On the left side of the screen you can see the overall waveform of the sample, in this case a violin. In the time ruler above the waveform you can see a green bar labeled Sustaining Loop. This is the portion of the sample that is looped. On the right side of the screen you can see a close up view of the loop point. The left half of the wave is the end of the sample, and the right part of the wave is the start of the loop point. The trick here is to line up the loop points so the two parts intersect with the zero amplitude cross point. This way you avoid any clicks or pops that might be introduced when the sampler starts looping the playback.The first bit of metadata you need to add to each sample is a loop start and loop end marker. Because you’re working with prerecorded sounds, the sound doesn’t necessarily keep playing just because you’re still holding the key down on the keyboard. You could just record your samples so they hold on for a long time, but that would use up an unnecessary amount of RAM. Instead, you can tell the sampler to play the file from the beginning and stop playing the file when the key on the keyboard is released. If the key is still down when the sampler reaches the end of the file, the sampler can start playing a small portion of the sample over and over in an endless loop until the note is released. The challenge here is finding a portion of the sample that loops naturally without any clicks or other swells in amplitude or harmonics.

In some sample editors you can also add other metadata that saves you programming time later. For WAV and AIF files, you can add information about the root pitch of the sample and the range of notes this sample should cover. You can also add information about the loop behavior. For example, do you want the sample to continue to loop during the release of the amplitude envelope, or do you want it to start playing through to the end of the file? You could set the sample not to loop at all and instead play as a “one shot” sample. This means the sample ignores Note Off events and play the sample from beginning to end every time. Some samplers can read that metadata and do some pre-programming for you on the sampler when you load the sample.

Once the samples are ready, you can load them into the sampler. If you weren’t able to add the metadata about root key and loop type, you’ll have to add that manually into the sampler for each sample. You’ll also need to decide which notes trigger each sample and which velocities each sample responds to. This process of assigning root keys and key ranges to all your samples is a time consuming but essential process. Figure 6.39 shows a list of sample WAV files loaded in a software sampler. In the figure we have the sample “PianoC43.wav” selected. You can see in the center of the screen the span of keys that have been assigned to that sample. Along the bottom row of the screen you can see the root key, loop, and velocity assignments for that sample.

Once you have all your samples loaded and assigned, each sample can be passed through a filter and amplifier which can in turn be modulated using envelopes, LFO, and MIDI controller commands. Most samplers let you group samples together into zones or keygroups allowing you to apply a single set of filters, envelopes, etc. This feature can save a lot of time in programming. Imagine programming all of those settings on each of 100 samples without losing track of how far you are in the process. Figure 6.40 shows all the common synthesizer objects being applied to the “PianoC43.wav” sample.