Start Here

Let us begin with a full featured generator. No shortcuts, no missing features, no baby talk. I took the time to link you to every word you might need to google, as well as a full explanation of why it exists and (eventually) how each knob sounds. This is your time to learn at your own pace with no judgement. Hell, you might even impress onlookers for learning applied physics on your off hours.

Phase is position. Phase occupies time, frequency, and stereo.
Begin with oscilator position. You may have a random position knob. Turn it off for the sake of this demonstration.

When I wheel the phase position, my cycle starts in a new spot. This is extremely advantageous in sound design.

1. When pairing 2 identical wave shapes, I can control the amplitude via phase cancelation.
2. When pairing 2 unique waveshapes, I can use phase cancelation to control the intensity of my bass.
3. When making a chord, I can limit the cycle randomness window to the weakest and/or strongest polarity.
4. When making an arp, the random window can help or hinder the organic quality of the patch.

Detune only activates when unison(multiple voicings) is active. Unison may work in percentages, but it is always measured in centimeters. For even voicings, each cent is the difference in pitch between each voice. Odd voicings retain a singular mono cycle that stays in the pitch of your oscilator's tuning profile. Blend is an extension of detune which exclusively affects the stereo width of a sound. It is safest to leave a unison blend knob alone, or at 75% width, for most genres.

The pan knob controls absolute stereo position as a dry/wet feature. can be mapped to anything. Envelopes, LFOs, chaos algorithms, macros... the sound engineer is only limited by their scope.

Wave Shapes

have a single fundamental and 0 harmonics. In the real world, sines project a perfect circle. On a 2D plane, they make a parabola.

add subtle harmonic content due to the linear sequencing of the waveform. you receive more highs than you would from a sine, but less than a saw. triangles were used in soundcards with low sample rates to mimic a sine, but have an equally important place in analog history.

lose high end at their peak. In other words, the volume of a saw wave's harmonics decrease in volume in a linear manner. This rounding of the top end is caused by an exponential loss of amplitude, much like you would see in a low pass filter.

only move up and down on a 2D plane, but they take the shape of a square when played in the real world. They have a near infinitely repeating harmonic spectrum which is best used with filters the cut the high end.

is a form of wave shape manipulation that takes the beginning of a wave cycle and creates a linear fold, much like one would see in a square. There are few sounds like a pulse wave, but should be saved for use in deliberate situations. Where you gain high end, you will lose mids.

Wave-shapes may be tested using a speaker driver + any of the following:

1. - a metal plate and sand
2. - a glass of water
3. - a really big speaker and a slow motion camera

For those on a budget, it would be more practical to install a stereo imager and set two voices in unison without random and low detune.

Synthesis

is the process of combining 2 signals together to create a new waveshape. Traditional additive synthesis is the result of stacking sines. If you only use odd intervals, you can create a triangle or saw. An even set of intervals can produce a square. These waveshapes including a sine are called Basic Shapes because they are the 4 most commonly produced waveforms using additive synthesis.

1. Saw waves are even and odd harmonics.
2. Square waves harmonics are odd only.

is functionally filtering, or EQ. Filtering is what makes most synthetic instruments sound organic, "wet", or otherwise interesting to listeners. Phase cancelation also applies to subtractive synthesis.

1. Phase Cancelation:
   Take two oscilators of any kind. These two outputs will phase with eachother, removing inversions and amplifying overlapping tones.
2. Filtering:
   A filter will not create audio that is not present in the waveform. Lets assume we know our wave shapes. A negative (-) phase comb begins with a reductive low shelf and folds into a positive peak. This removes bass relative to the first fundamental, or peak, of the comb causing a metalic feedback. A positive (+) comb would then add bass via low shelf, remove equal volume via the first peak, and add on the second peak. This makes sounds squishy.

is a sample based method of audio manipulation, a.k.a. wave manipulation. Granular first requires a source file to scan through. Things that separate granular from sampling are amplitude gating and wave shape modulation, oscilation, grain - which allows for the chunking of audio in time, grain randomness in time, pitch and stereo, ,

build on the idea behind "Sample and Hold". Static oscilation of a source sample creates a sound similar to common waveforms oscillation. The magic behind spectral is user-made comb filtering, allowing for fine tuned resonances and hyper tonal pitch gating. Imagine a suite of fx dedicated to the degredation of bitrate and channeling of the frequency spectrum, with backward and forward control of a sample, much like a wavetable.

are seen in most modern soft-synths. Layering two waveshapes changes your output waveform via subtractive synthesis.

Warp Types

takes a waveform's cross point and displaces the waveform right and left of the vector in equal amounts.

takes a wave's x-point and displaces the x-point right or left of the vector in proportional amounts.

Inverts the source oscilator's waveform cycle and replaces it along the warp position's x-point

Combines Flip and Asymetrical. The waveform is inverted along the x axis and doubled in pitch (allows for manipulation of both sides of the cycle without breaking waveform). The x-point at 90° and 270° respective utilize asymetrical transform.

Pulse Wave Modulation adds a 0 point to the waveform, muting the affected cycle and squeezing the source waveform to compensate. PWM is one piece of what makes classic game music sound like video games.

is equal to downsampling. This warp reduces the number of samples in a logarithmic fashion. Reducing the sample rate introduces noise. In the case of downsampling, that noise is squares.

Modulation

Amplitude modulation turns the input signal into a volume LFO. The volume effect becomes more dramatic at lower frequencies (sub 30hz) and will otherwise provide an accompanying pitch and timbre which reflects the input modulation at higher frequencies.

Frequency modulation takes an input signal and vibrates the carrier oscilator according to the shape, or path, of the input signal.

Ring modulation feeds sounds to a waveshape and alters their sound based on the frequency of the ring mod. RM is altered by the source waveshapen the context of 2 oscilator set ups will recieve

Time Based FX

is the foundation of all 4 effects listed here.

Delay repeats audio as few as 2 times and up to infinitely as many times as the user sets on the feedback parameter. This effect originates with tape-based recording. Tape echo would lose high end so delay would become associated with a tapering in the high end over so many repeats of the source audio. The introduction of digital repeat allows for the inverted low end taper, meant to emulate the echo of a narrowing chamber, as well as create infinite repeats.

The collection of tools above are what are most commonly associated with delay, and will be the best use case for this effect type.

Delay is prefered over reverb on recordings of vocals in favor of reverb because reverb convolves will increase in volume as the length of signal grows. Drums alternatively get a huge buff from reverb due to their percussive nature.

creates tails. Compression can make reverb ugly but it can also make reverb larger than life.

takes an input signal and creates stereo copies. Each copy will recieve a time and pitch value, as well as a filter. The filtering smooths artifacts while the pitch modulation prevents the delayed copies from sounding like slapback. A chorus often ranges from 15-35 milliseconds, but there are digital and analog units that can reach 0ms.

is achieved through a number of mediums. To read how they are done, visit Stereo & Width

Delay based stereo effects are haas and "ping-pong" delay. Haas is a static, offset delay, while ping pong delay modulates between hard left and hard right panning.

Phase Based FX

EQ, or spectral equalization, has its own section called Subtractive Synthesis.

For information on parameters and their uses, .

The fx unit a producer chooses will affect the appreciation and respect they have for phasers. Phasers use notch filters called poles. Each pole sweeps the source output for overtones to boost.

A negative polarity can sound metalic because the phase begins with a peak. Begining with a peak means that the bass is shelved at negative volume. A positive polarity can sound wet or glassy. Rate will control the stereo position of each pole. Setting a phaser's rate to ±0hz will allow for static combing of the input's spectral range.

A flanger delays the signal to get a gooey mess.

Filtering

A phase filter

This filter is theoretically infinite. So long as there is signal to emphasize, a flanger can sweep it.

You can tell this is a comb filter because of the way that it is. Each notch is set by the user

Routing

should be handled with widening first, compression second, and distortion last. Group stacatto sounds that would share transients for optimal headroom. It is better to clip the same distortion at once than it is to have two sounds clipping one cieling out of turn. This offset clipping will harm the spectral dymaics of legato sounds and weaken the transients of a group's shared instruments as it pertains to the clipper/limiter's ADSR.

There are two ways to tackle stereo.
Unison and phase based widening should always come first. Putting these first removes the opportunity for human error. "What human error?", you might ask. Wideners do exactly what we tell them to. When audio is boosted in any capacity, the stereo may bleed out of the audible stereo field after high gain is achieved. Widening for stereo depth early on will prevent this by encapsulating the stereo data within the safe volume range.

Reverb and Delay are the tricky ones. We put these after compression and before post EQ to prevent an artificial infinite decay.

My prefered way to EQ is once before compression and once before limiting to max loudness. These two EQs will be refered to as Pre EQ and Post EQ. The Master EQ will handle the absolute highest and lowest frequencies that reach the final render.

Pre EQ should be used to tell the compressor what to amplify. Boost frequencies in this instance. When there is too much bass, use a shelf. Highs add a human touch to the sound so only bring a lowpass into the mix if the high end goes above the average volume.

Post EQ should only reduce frequencies that are clipping. Cut frequencies in this. For max headroom, cut bass on most sounds at the 2nd octave of the root note. Highs can be cut with a 6db slope after 10khz. Most sounds work at the 9th octave, but others may not need the highs removed at all.

is the oomph behind a bass, and may be used to gain-stage vocals as well. The first part to consider is waveshaping. Want girth? Choose a sine-like wave. Want crunch? Square things out either through high input gain or a square-like wave. Need transparency? Use a simple linear clip, or a distortion that emulates an LED diode.

The goal with saturating last is to shave the top of your transients in a way that sounds natural. A limiter is more than capable of doing the job, but limiters only kill transients with downward compression. the addition of waveshaping and multiband distortion provides an opportunity for more gain.

Layering

Layering sounds is not a new idea in music production, but its far more necessary when synthesizing instruments. A preset made by a radio quality artist may not require layers. A single full bodied patch especially does not need to be "frequency split" or any other gaudy trend hopper gimmicks. Does it do what its expected to do? leave it.

Lets start with the human voice. A mouth is a white noise oscilator, a fourier filter, a variable width bandpass, and an envelope. Most music we listen to in the western world is vocal ventric. For dubstep, our layering for a main bass is dependent on meeting all 4 criteria. A filtered saw wave can achieve the required timbre for most mid range voicing, while a noice osc can give a synth breath. This then opens the door for consonants. Our first layer can short hits from a sampler that sync up to the attack on our midi notes. Maybe some harshness is needed to emulate a scream vocal. Record yourself scrubbing wood with a rasp and throw it underneath the sustain.

Percussive sounds need a little more. Listeners have grown acustom to decades of room recordings of orchestras. Drums do not sound like drums until air is introduced. When that air reflects off the room, it adds decay, echo, bass, cross phasing(in stereo), and mechanical noise from the hardware used to amplify and sculpt the sounds from the membranes and tubes of the percussion. Our job is to mimic or even invent the idea of a tonal pluck.

So lets say you are responsible for inventing this tonal pluck.
Draw from reality. A car door slams. The pitch raises at the transient. A snare cracks. The pitch, again, raises at the transient. As these sounds decay, they drop in pitch and then shortly after, they hold or lose their pitch. A lot of work is in the pitch modulation. The transient, or the snap, is easily handled with a literal snap sfx. This may be a snare sample or a clip of white noise that has been cut down to the shortest sound.

Remember - layering is a tool, not a necessity.
Lets pretend your synth cannot provide the crack you need to make the sound hit. Try adding a transient hit to the snare group so the bass can snap into the groove when its needed. Maybe the bass is too flimsy. Group the main bass and an audio channel and add a shout where the stab's sustain would occur, without removing either the stab's sustain or the chant's sustain. your snare could need some life. Give this snare a pitch altering layer in the same was as the bass stab.

Transients

A waveform has 4 parts. Attack, Delay, Sustain, and Release. The transient is the node where Attack and Delay meet.

An amplitude transient is the loudest part and nothing but the loudest part.

A pitch transient is the part which may resemble a pop or thud.

are most commonly associated with drums. Amplitude(volume) transients can be used to enhance the snappiness of an instrument without taking from available headroom. This is best applied to plucks, stabs, and any synthesized drums.

can assist where volume fails to punch a hole in the mix. A pitch transient is commonly used to make a kick drum or 808. What makes pitch unique is that squeak or thunk that comes from

Resonance

Resonance has a few vocabulary terms to know. Common ones are bias, ratio, peak, has its own section called Phase.

For information on parameters and their uses, .

Combs will have a range of peaks and troughs. A notch, bell, and cutoff will have 1. Shelves and bands have 2 and 3 respectively.

comes from too much resonance. This may be utilized to one's advantage with enough practice. Feedback is treated as sustain in things like resonators. Feedback is tamed with gates or modulation when used in subtractive synthesis (filtering).

is the tuning of the filter, or the resonant frequencie(s).