Hume provides a completely unique system for generating gorgeous note sweeps and subtly shifting tones.
Here is a video sample. You may pop up the video window for full-size video (1280x704 pixels).
I got as far as adding the Synthcore 5D filter to Hume before people started complaining about my work in Reaktor again. Here is a screenshot of the design.
The filter can continually change cutoff, resonance, drive, filter type, and poles.
- The filter is 3x oversampled, providing consistent performance up to the C8 octave at standard sampling rates.
- The resonance can increase until there is only self oscillation. the filter is fully gain compensated, so it won't shatter your eardrums.
- At lower pole settings, the source signal (oscillators in this case) mixes with a 2-pole filter. At midpoint, there is only two-pole filtering. At higher pole settings, a 4-pole mixes with the 2-pole. At maximum, there is only a 4-pole filter.<.li>
At lower drive settings, the filter resonance dominates the output without distortion. As drive increases, the resonance diminishes compared to the source and filtered signal. As drive increases to the midpoint, the filter itself saturates. At higher drive settings, the output also saturates, including any source signal, until it fully clips.
- The type can also be smoothly changed between low, band, and high pass.
- The polyphonic display shows the filter curve for each voice, with each voice in a different color, after gain adjustment by the amplitude envelope.
- Over the display, you can click and drag the mouse to change the cutoff (horizontal) and resonance (vertical) with the LEFT mouse button. with the RIGHT mouse button, you can click and drag to change the filter type (horizontal) and poles (vertical).; The slider bars around the XY control for the filter let you zoom in on a particular area for finer mouse control. You can also set values specifically in the number boxes on the top right of the filter area by setting focus on them and entering numbers, or by clicking focus onto the number boxes and dragging up and down.
- Four modulations are available to any of the five filter dimensions. There are 40 sources and each may be scaled separately.
1. Hume3 for R6
Hume's unique sampling sequencer and morphing oscillators create complex, delicate, and rich timbres, both in sonorous pads and in complex rhythmic patterns.
The instrument has three main sections: sound generators, sampling sequencers, audio post processing.
2. Architecture Overview
2.1. Sound Generators
With an XY panel, you can adjust the mix of 4 oscillator sources. There is one oscillator in each corner--the center of the panel is an even mix of all four.
Around the XY panel, morph controls can modulate the mix from dozens of modulation sources, and you can see the resulting waveform in the panel at all times. When the morph sliders are centered, the sound is constant. When moved to the left or right, the morph modulator acts like vector mixing.
On the LEFT of the morph control are two width-modulated oscillators, with up to 4 waveforms each, and patchable modulation sources for the width and pitch.
On the RIGHT are two FM oscillators with up to 5 waveforms each, and patchable FM source, AM source, and pitch mod source. For each oscillator you can make one waveform audible, and also use any oscillator waveform as an AM or FM sources. So the oscillators by themselves are capable of alot of different sounds--for example, you can use a osc1's pulse oscillator in the sound mix, and also use osc1's triangle oscillator as an FM modulation source for oscillator 3.
The morphed output is mixed with a separate noise generator (with its own amplitude and color modulation); then vibrato and tremolo may be applied to the entire sound from any modulation source. Polyphonic tempo glide, glissando, tremolo, and vibrato are available. Each snap can spread the amount of pitch detuning across the voices (voice pitch spread) by a different amount.
This design is different from other sequencers in a special way. This is, the sequencer itself doesn't generate triggers or gates. Instead, there are two envelopes which *sample* the sequencer and put out a note with whatever pitch and velocity the sequencer has at that point in time. There is a also separate pitch transposer that works like an arpeggiator.
The envelopes, sequencer, and transposer all run independently of each other, each with a DIV control that divides down the trigger source. For example, if an envelope DIV is 2 and the sequencer DIV is 4, then the envelope plays each note from the sequencer twice. If it's the other way around--the DIV is higher for the envelope than the sequencer--then the envelope skips notes from the sequencer. By setting uneven div rations, you can create all sorts of patterns. In addition, there are two trigger sources for the sequencer, transposer, and two envelopes: a shuffle clock and beat sequencer. The shuffle clock can use a switchable internal/external clock source and adds swing to odd or even triggers. The shuffle clock directly runs the beat sequencer.
The beat sequencer is a simple cyclic on/off trigger for step. It generates a gate length depending on the number of OFF steps following each ON step. The beat sequencer trigger can be divided down from the clock source, and it can have any number of steps. Either the beat sequencer or clock can trigger the pattern sequencer. The pattern sequencer has up to 8 steps, with pitch and velocity controls for each. The pattern sequencer can run at various divider ratios of the shuffle clock, or on high triggers from the beat sequencer. It can run forwards, backwards, and in two reversible modes.
The sequencer pitch is fed into a transposer which can transpose the 4 notes at variable intervals, over variable ranges, in various directions. The interval is again a settable division of the shuffle clock or high triggers from the beat sequencer. The sequencer pitch passes though a pitch remapping unit, which can optionally force notes to a particular scale, and microtuning, before it's sent to the envelope samplers.
When an envelope samples pitch and velocity data, it sends the resulting pitch and gate information out as MIDI data, which is routed back into the same instrument. So you can also play notes from a MIDI keyboard, for example, even when the sequence is running.
2.3. Post Processing
The oscillators feed a parametric saturated filter, designed for low CPU usage while offering rich harmonics. More than a dozen filter modes are available. Patchable modulation sources can control frequency, Q, and saturation. The saturation and Q response curves are shaped for highly resonant timbres. The filter output is summed to a monophonic source, then feeds a mono/poly compressor and reverb chain. The reverb's mono output is moved around in a stereo field using pan modulation, which then feeds a stereo echo with tempo and modulation delay.
3. Panel Objects
3.1. Modulation Matrix
Modulation sources include:
- Three LFOs with shape, staircase, sync, and AM modulation controls.
- Two ADSR envelopes which may be triggered both by the pattern sequencer and incoming notes. Velocity can modify the output level, attack, and (for sequenced notes) gate width. A gate divider circuit can control the frequency at which the envelopes are triggered and (on sequenced notes) adjust the gate width.
- A compound matrix modifier provides mixing of continuous and real-time modulation sources.
The delay lines can also act as a chorus stage which --in combination with the morphing sound source--give you scintillating, wide, fat, pads for novel sonic soundscapes.
The overall result is an endlessly changing, morphing waveform source, capable of both analog-style and FM-style sound generation in an intuitive interface. After sonic and dynamics enhancement by the filter, compressor, and reverb, infinitely changing sounds are layered over themselves by the pattern sequencer and tempo delay in exciting rhythmic patterns, either pure or mean, and it's all easy to adjust dynamically. Intended for real-time tweaking and fun.
All LFOs have polyphonic spread. When SPREAD is set to 0 (and sync is off), each LFO is monophonic. At higher SPREAD values, a phase difference is added to each LFO voice, and each voice's frequency is spread out slightly, providing a phatter sound.
The meter in each LFO shows the output level of all voices, instead of just one voice.
The LFO modulator can effect frequency or amplitude level for each LFO. The LFO mod source and sync are controlled by event logic, instead of switches, so changing their settings does not interrupt the sound.
Voice morph modulation is polyphonic. The current morph position for each voice is displayed in the thin strips around the oscilloscope. In the bottom right, SPREAD is now available for oscillators voices. This provides fine detuning of the synth voices from each other. If set to the maximum of 0.5, then the voices are detuned by 500 cents. A setting of 0.1 or less provides a tiny amount of detuning (for that phat sound).
40 modulation sources are available for each oscillator's AM and FM source. An oscillator can modulate itself (with its current or another waveform). AM is available for osc 1. Noise can be switched off, to reduce CPU usage.
BIAS, just to left of filter, adjusts the slope of filter response around the chosen center pitch. This is quite a dramatic control, often well affected by real time control by mod wheel or pitch bend.
When morph modulations exceed the -1/+1 range, they are mirrored into range.
Polyphonic pitch glide causes the pitch for each new note to slide from the pitch of the last voice. Tempo may be taken from the clock or beat sequencer. For slower glide, user lower clock rates.
- The compressor has Polyphonic as well as monophonic mode. In poly mode, each voice is compressed separately, so transient peaks in each voice do not affect other voices.
- The reverb has "higher cpu mode" as well as original "lower cpu mode". In higher cpu mode, an additional feedback loop thickens the reverb tail and the mix level is controlled at output, rather than input, so thick reverb sound can be mixed in at a low level with the original signal.
- In the echo unit, internal switching removes pitch glitch (due to delay time changes) when delay time is not being modulated. Its mixer and audio path is also designed to reduce cpu and reduce gain swell when changing snaps.
- Modulation of filter saturation (by MS1 and MS2) controls the amount of modulation by source M1 (the upper list box in the filter's bottom left corner) and MS2 controls modulation by M2 (the lower listbox).
- The beat sequencer has its own clock divider and length setting, like the pattern sequencer.
- The beat sequencer issues the set number of rest periods after each ON step, which is multiplied by the clock period to provide the actual duration before the next ON step.
- Note generation is designed to stop stuck notes on snap changes. The on-screen keyboard, envelope, and MIDI automatically turn off all playing notes when changing snaps.
- Reset logic is cascaded. The instrument attempts to reset all clock dividers in all sequencer and envelopes properly on clock pause and snap change.
The HOLD and PREDELAY settings in the envelopes are set as multiples of the clock or beat sequencer duration. If set to 1, and not otherwise modulated, the period is the same as the length of the current note. The maximum hold time and predelay range is settable to:
(multiples of current note duration) * DIV * (number of voices)
This lets you choose the maximum possible length for each voice when the gate source is the clock (whether divided or not) . Longer notes are not possible (because then more than the number of voices available in the instrument would need to play).
- If using the beat sequencer as the gate source, and there are many OFF steps in the beat sequencer, be careful to use lower settings for HOLD and PREDELAY or there could be some hung notes. Of course sometimes that sounds quite good. (Changing the snap or pausing the clock clears stuck notes).
- You can always increase the number of voices in the instrument (cpu is low!)
- If SEQ is off, the DIV control divides the number of incoming gate triggers (from MIDI or triggers generated by the other envelope). This is useful for example if Env1 is playing a sequence with DIV 1, and then ENV2 can have DIV set to 2 so the envelope only triggers for every other note. If not using the sequencer (playing from a MIDI keyboard), usually you would set DIV to 1 so each midi note plays--or you can use it as an accompaniment that skips notes that you play, by setting DIV to a higher value.
- The UNISON and MONO modes are designed to support the above features properly. For example, if env2 has SEQ off, DIV set to 2, and MONO on, then every other note from env1 triggers the Env2 envelope for every voice.
- The envelope gate modulation source has been changed to envelope velocity.
At request of a competitor I was banned from NI. So now you may find the files aton this site.
Thank you for your interest in Hume. I hope you enjoy it!
-Ernest Leonardo Meyer