Look at any waveform and you can easily break it down into a couple simple parts. The spike at the beginning of the sound is the Transient. About 10-20ms after the transient comes the rest of the sound, the Sustain, or the actual tone of the instrument or voice.
Just about every recorded sound imaginable has these characteristics; it’s pretty rare to see a waveform from an acoustic instrument that lacks a Transient or Sustain. So it is safe to say we are discussing 99% of the waveforms you are going to encounter when working with audio.
TRANSIENTS
When a pick hits a guitar string, a stick hits a drum, or a bow begins to vibrate a cello string, there is a distinct and very audible, non-tonal noise that precedes the tone it was meant to generate. I say noise here because it may be beautiful in the most musical sense, but it is probably not the sole purpose of the instrument. Even a Pizzicato pluck on a violin provides some tone, albeit brief, otherwise you wouldn’t be able to tell what note was plucked.
Transients usually have both a non-tonal characteristic and a very brief lifespan. For a snare, it’s about 20ms. That’s hardly measurable to the ear. Remove the first 10-20ms of a snare and you end-up with more of a blob of sound. No crack, no top mic boom.
It is safe to say the Transient is both a critical part of sound, and one that invokes a ton of engineering. For example, take the process of mixing compression, which was invented to even-out the presentation of audio. Compression is a complex and frequently used technique employed by mixing engineers in order to manage the spikes of Transients. There are a lot of moving parts in compression, and it can be a very confusing topic for aspiring audio engineers. But if you aren’t compressing your tracks, correctly, you run the risk of allowing peaks from your instrument or source to override the tonal qualities of the sound.
And let me be clear here, a Transient isn’t just the beginning of a waveform. A violin playing an arpeggio of notes will have only one starting point. In between each new note or string you will likely see a new Transient. Even vibrato of a single, held note can result in Transients. Everything between these Transients for me is Sustain.
TRANSIENT NOISE
Do Transients really not have tone? Of course they do, but they are presented differently than Sustain. If a Transient didn’t have a tone, we couldn’t hear it. After all, it has to occupy somewhere between 20-20khz to be audible by humans. But you simply cannot say the stick hitting a floor tom is the same as the ring of the tom that results from that kinetic energy. They are two very different, albeit dependent, things. Which is why I refer to Transients as more non-tonal noise than anything else. Take-out the first 5-10ms of a single sung word or that tom hit, it will likely not sound too musical.
If you generate a single sine wave with a Test Oscillator, and put an EQ with a spectrum analyzer into the chain after it, no matter the frequency you programmed into the oscillator, you will see a single spike on the frequency spectrum. That is the “Fundamental” frequency of that sound. The tone of the instrument. If you analyze the transient frequencies generated by a stick hitting a snare rim or a bow scraping a string, it may be difficult to identify any fundamental frequency. Instead, you will likely see a lot of different frequencies generated. No single spike in a specific range but a bunch of spikes spread across the 20-20k range.
NON-TRANSIENT NOISE
There are a lot of words used to describe the non-transient parts of a waveform, so I am not going to get into semantics. Let’s just agree if it’s not a Transient, it is something else. But the same non-fundamental noise can be found within the non-transient parts of a waveform. Some of this is part of the natural instrument itself, and some of it is captured through the recording process or by adding saturation.
If you take the sine wave we generated earlier and add a Decapitator - my favorite saturation plugin - before the EQ in the chain, and then analyze the frequencies generated, you will now see more than a single frequency spike. And while it does not really change the fundamental tone of that original frequency, it just sound bigger. This is because the saturator has added harmonic content to the original sound. Usually this is generated by adding frequencies that are the product of odd or even numbers multiplied by the fundamental.
Take that sine wave generated at 100hz. If you were to duplicate that frequency at 200, 400, 800hz and so on, you would find your original 100hz note now starts to sound like a chord on a piano. And in fact, that is exactly what is happening. Hit a C1 on a piano, then add C2, C3, and C4. Played simultaneously, you have now have a much more rich representation of the fundamental C1.
When you plug a guitar into a tube amplifier, the tubes will generate an even harmonic series. Multiplying the fundamental by 2, 4, 6, and so on. Tube amps can make a ho-him guitar really sing with mellow, warm overtones. This is why tube amps are so coveted.
Take that same guitar and plug it into a solid state amp, and odd harmonics will be generated. Some would describe this tone as crisp or edgy. This is because odd and even harmomics are very different when it comes to adding them to a fundamental. To add odd harmonics to a fundamental tone, simply multiply the fundamental by an odd number series. A 100hz tone may have additional odd harmomics at the 3x, 5x, 7x, 9x, and so on frequencies (e.g., 300hz, 500hz, 700hz, etc.).
Depending on the type of saturation you are employing, you may say the additional frequencies are adding noise. And yes, noise typically elicits a negative translation, but in the audio engineering world, noise can be musical. Could you imagine a drum track that lacked all non-tonal transients? Pretty boring. Which brings us to the rest of the waveform, the part that invokes the fundamental frequency, Sustain.
SUSTAIN
There are very few acoustic instruments that lack sustain. Even wood blocks generate a tone. Regardless how long or short the Sustain, there is really no sound with sustain that is without tone. Even a chainsaw has tone, but it is also very noisy which means it isn’t just a single fundamental tone being generated, but a bunch of frequencies, many of them in the odd and even harmonic series.
Simply put, the kinetic energy exerted during the Transient creates a vibration that Sustains for a period of time. And that vibration whether it is a single note or a range of frequencies, is Sustain.
So it is safe to say while Sustain may be the product of a Transient, especially on an acoustic instrument, without it, we would likely have a lot of 20ms clicks and scratches.
Depending on the instrument, however, sustain may be a big deal. A guitarist may be picky - pun intended - about the loudness of his picking compared to the sustain of his strings; the chug of his guitar compared to open chords. There is always a balance between Transient and Sustain that must be respected.
For vocalists, of course managing plosives, clicks of the tongue, and breaths is a huge issue. Heck, even breathing generates Transients. But Sustain tends to dominate the workload for vocals. Once you get beyond micing challenges, it is safe to say there is not a lot of non-tonal energy to a singer’s voice, so Transients are less of an issue.
I spend much of my vocal mixing time managing the impact of transients and the tone of Sustain. Recorded voices tend to be very strong between 1-2khz, especially from a powerful singer. this is also the sweet spot for a vocal in a mix. Too much and the listeners ear will become exhausted quickly; tends to be perceived as non-musical at some point. Too little, and the vocalist will be drowned-out by the mix. Allowing a little of the transient in the 1-2khz range to poke through, I can retain clarity. But then after about 5-10ms, I compress that range by 3-6dB with a Dynamic EQ which keeps it musical. In essence, I am allowing the Transient of the vocal to punch through the mix and then controlling the presentation of the Sustain in order to sit the vocal back into the mix.
SUSTAIN SATURATION
Since the Sustain part of the waveform is much more tonal, and the Transient can be less musical, when I apply saturation to a recoding, I try to limit it to the Sustain portion of the waveform. Adding harmonic content to the entire waveform also enhances the Transient, the non-tonal qualities of a waveform. You usually don’t need to amplify non-tonal noise like a pick or stick, you can do that with Sustain Compression (see Compressors Demystified). But if you want to enhance the ring of a guitar string or a floor tom, it takes a little wizardry to separate the waveform from its Transient before you send it through a saturator.
If you reference my work in Compressors Demystified, what we are talking about here is Transient Compression. We want to basically remove the Transient as much as possible from the waveform by employing a very fast attack compressor. Once the transient has been tamed, follow the compressor with a saturator. Again, I prefer the Decapitator, but I would also recommend using Scheps Parallel Particles, Infected Mushroom Pusher, Saphira, SDRR, Logic Pro PhatFX, or anything else that can add harmonic content. It should be said here Saturation is not a one tool fits all technique. Sometimes I have to try three of four of them to get what I want.
FINAL NOTE
One final note regarding Transients and Sustain. When is comes to mix engineering, there is no hard and fast rule regarding when a Transient ends and Sustain begins. For some instruments this can be a very fluid ratio, especially for vocals.
Eventide invented an EQ that allows you to separate the Transient from Sustain using a millisecond timer. It’scalled Split EQ. A key component to applying this EQ to each part of the waveform is understanding how many milliseconds to allow for the Transient until it hands-over control to Sustain. And yes, there are presets for most instruments like drums and vocals which is a good start, but I tend to rely on my ears more than anything else to ensure the control I am exerting over a Transient or Sustain is giving me what I am looking to achieve.
The same goes for compression or any other mix engineering that involves Transient or Sustain management. Before I attempt to modify a waveform, I ask myself what am I looking to accomplish, and in my head I can already hear it. This is not a mental calculation, nor do I say to myself “shave 5-10dB off the first 10ms of a transient”. I use my ears to determine if the change is desirable. For every instrument, every track, and every waveform, this is a very subjective and non-mathematical process.
Comments