Audio or MIDI? That is the question. Jono Buchanan looks into the importance of understanding the inherent strengths and weaknesses of both, and the many ways to extend the relationship between them.
Nearly all DAWs offer the chance to work with two types of data, region or block, as nearly all provide both MIDI and audio capabilities these days. Whether you’re looking to fire up an instrument and doodle away with your controller keyboard until creativity strikes, or recording, chopping up and editing “real” sounds onto audio tracks, these two data types can happily exist side-by-side within one arrangement.
However, through this article, we’re going to look at ways to extend the relationship between these data types further, exploring techniques which cross from MIDI to audio and vice versa. The ways in which we can interact with audio differ from those in which we can with MIDI. This suggests that the editing processes for each will also be different. By learning what these differences are and how they can benefit your work in progress, it’s easy start thinking along the lines of “if this MIDI part was audio, I’d be able to…” or “if this audio file was mapped to my MIDI keyboard, I could…” Switching back and forth, converting one data type to another so that techniques can be applied more creatively is a weapon top producers use every day.
Converting to audio
Let’s start by looking at some examples of when converting MIDI parts you’ve programmed for synths, drums or other plug-in instruments might benefit from being bounced down as audio files. From a purely administrative point of view, the first benefit comes if you’re sharing the parts of your track with a co-writer or remixer who is working in a different DAW or with different plug-ins. To ensure an absence of compatibility issues, bouncing stem mixes of groups of files or, better still, individual tracks with each part rendered as audio, means that DAW-swapping problems are a thing of the past. However, the benefits of bouncing MIDI to audio goes much further.
For all of MIDI’s strengths, there are some issues which can’t be resolved for a part that is MIDI-based, while other tasks are much more long-winded than they would be were the data in audio form. For instance, if you want to reverse a sound, there’s no way to do that with MIDI, but once a part is bounced to audio, it’s a simple task to find the “reverse” function and apply it to your audio file.
Reversing is effective not only when it’s applied to entire audio files but also when reversed reverb is employed to preempt a sound. Normally, reverb provides a decay tail after a sound to create the illusion that the note decays into an acoustic space but, what if that space somehow “knew” the sound was coming and faded up in volume to meet it, rather than providing a tail as normal? This is the trick made famous in “Personal Jesus” by Depeche Mode, where the “reach out, touch faith” line is pre-empted by a reverb rush which rises to meet the vocal line. Reverse reverb tricks are achieved firstly by bouncing the dry file you want to apply reverb to down as an audio file.
Gating without Noise Gates
Audio files are great for more than just producing reversing tricks. In particular, they can work wonderfully well for creating quasi-gating effects, which are particularly useful for pad sounds with slow attack times. If you have a MIDI pad part that are changing chords every bar but which takes a little while to attack, there’s no way you can use MIDI to create a busy sequence part, as whenever you try to repeatedly trigger the sound, you’ll find it doesn’t attack quickly enough.
One way around this is to bounce the pad part down as an audio file and then use the “chopping” tool within your DAW to cut the part into slices. If these slices lock to the 8th or 16th notes of your track, you can create rhythms. However, this won’t sound effective until you’ve selected the chopped files and dragged their end points “backwards” to create short gaps between the regions. This will provide a blast of audio followed by a short period of silence, providing a gated effect. Often, these chops will sound unpleasantly “glitchy,” as the audio regions will click every time they start and stop. Most DAWs allow you to create a short real-time fade in/out for each region and if you apply these fades, with short values such as two or three milliseconds, you’ll get rid of the clicks without affecting the percussive nature of each slice of audio.
Less dramatic, but just as effective, is the notion of being able to stop long release times once a file is bounced as audio. If you’re using a pad which has a long release and you come to a point in your track where you want an abrupt change of mood, it can be frustrating to have to wait for the pad to decay fully, as this can dilute the effect of the track changing character. By bouncing the pad down, it’s a simple job to chop it at the point where you want its sound to stop, creating a much clearer division between one section and the next. It’s what I call a “jigsaw piece edit,” as you can see and hear the line where a sound stops so clearly. If it’s too much to have the sound stop completely dead at the chop point, extend its region a little and apply another real-time fade, so that you’re compromising the difference between a full “hard” and more gentle “soft” edit.
Reversing the MIDI to audio concept has its own advantages too. The most effective way to do this is to use sampling techniques, which will allow you to take any audio file and convert it to MIDI so that it can be triggered, either as one long file, or in slices. Looking at the advantage of using one file first, for beat programming in particular, it can be great to layer up loops of original, double, half and quarter-time lengths.
You might be wondering why this works, so here’s a short explanation. Samplers capture a sound using a sample frequency, in the same way that pitch is also measured in frequency. In pitch frequency terms, let’s take the note A above middle C, which has a frequency of 440Hz. The A at an octave below has a frequency of 220Hz and the one below that, 110Hz. The reason As are all called “A” is because this mathematical link exists between them—for each octave rise, the frequency will double, and for each octave drop, it halves. So it is with samplers.
More often than not, sampling frequency is set at 44.1kHz, so by triggering a note an octave below its original sampling time/pitch, the playback frequency is 22.05kHz, producing a halving of pitch and time.
However, the disadvantage of long blocks of sound triggered from a single key becomes all too obvious if you suddenly change the tempo of your track. Suddenly, rather than having a loop which sits comfortably with the rest of your production, you’ll suddenly discover that, as the loop is simply triggered at the beginning of a bar but then has no tempo reference points until it’s triggered again, it’ll be out of time.
As with the pads example given above, to rectify this, chop your audio file into slices, with either 16th note or 1/32nd note resolution. Then save each slice as a new audio file. Most DAWs allow you to select lots of audio regions and save them as separate files in a single operation, so it shouldn’t be too painful to make this happen. You can then drag the audio files into your sampler and assign each one to its own key—again, this is usually possible via a single operation so that the first file maps to C1, the second to C#1 and so on. All you need to do to have the original loop play back via MIDI is to draw a ramp of notes from C1 upwards in a semi-tone scale, with a note being triggered in the time-frame you selected when you sliced the original audio loop up. In other words, if you chopped in 16th notes, enter an event every 16th note.
When you press play, the loop should sound identical. Now, though, you can change playback speed. As each slice is being triggered from its own position within your track, if you slow down or speed up the tempo, the positions of those slices will still trigger where they’re stored, so while you might get brief gaps in the audio file, or a few glitches if you slow it down a great deal, the loop will still run in time.
Often, DAWs contain dedicated “Convert To Sampler” options which carry out the task of changing an audio file into a chopped sampler track automatically, saving you the time of the actions described above.
Lastly, here’s one final trick to try once your audio file is converted to your sampler and is being MIDI triggered. Adjust the amplifier envelope settings within your sampler, these are likely to have immediate attack, little or no decay, full sustain and little or no release, as these settings will play each “slice” to sound like the original audio. However, if you drop sustain to zero, you can use the decay time to control how long each slice will play for.
Published / Tuesday, 18 October 2011