Offline Synthesis / IR Guides

These guides match the Package Browser’s Offline Synthesis / IR group. They cover selected offline synthesis, resynthesis, grain-cloud, and impulse-oriented tools.

Dense Grain Cloud

Use this for source-based grain clouds rendered directly to a multichannel item.

Selection:

• Select one WAV-backed media item.
• Run Dense Grain Cloud.

Important controls:

• Grains sets the event count.
• Density admits more or fewer grains.
• Grain ms controls average grain duration.
• Length variation makes the cloud less uniform.
• Pitch scatter oct moves grains away from the source pitch.
• Spatial spread and Channel contrast control how strongly grains separate across channels.

Use lower grain counts while setting the event shape and spatial spread, then raise the count for denser renders. Use the amplitude envelope for fades instead of relying only on post-render gain.

EVP Field

Use this to render synthetic speech-like formant material as a multichannel or ambisonic field. eSpeak NG creates the spoken source, then the selected voice treatment reshapes it before spatial rendering. eSpeak NG is required for this process. Speex is optional and is only required when Codec damage is enabled.

Output:

• 3OA ACN/SN3D writes a 16-channel ambisonic item.
• Multichannel shape writes a speaker-shaped item using ring, double-ring, cube, or dome placement.

Main controls:

• Voice treatment selects clear speech, sung chant, melodic, choir, whisper ghost, possession, or broken radio behavior.
• eSpeak voice, Speech speed wpm, and Speech pitch control the source speech render. On macOS, install eSpeak NG with brew install espeak-ng.
• Time expansion controls how the spoken source occupies the render: Fill duration stretches the phrase to the requested duration, Manual expansion uses the expansion factor, and Speech length keeps the natural eSpeak phrase length.
• Scale, Root semitone, and Melody depth shape the pitch contour used by sung and melodic treatments.
• Vowel sustain, Choir voices, Shadow voice, and Ghost mix shape layered voice behavior.
• Spectral shaper adds an end-stage color mask before spatial output: Preset uses internal resonant shapes, Tone list builds a harmonic mask from typed note names, MIDI numbers, or frequencies, and Profile item imprints the spectrum of the first selected WAV media item.
• Shaper strength, Shaper bandwidth, Shaper partials, and Shaper brightness control how strongly the mask is applied.
• Codec damage runs the treated speech through Speex before spatial output. Narrowband, wideband, and ultra-wideband modes use different codec sample rates. Codec amount blends the damaged signal with the source, Codec quality sets the Speex encode quality, Packet loss simulates dropped codec packets, and Residue instead of decoded speech uses the difference between the source and codec result as the damaged layer.
• Pitch, Pitch spread, Formant shift, and Mouth size shift the vocal register and formant character.
• Breath, Noise, Whisper, and Formant smear add air, residue, and instability.
• Azimuth width, Elevation width, Spatial motion, and Speaker spread shape the spatial field.

Breakpoint curves can vary amplitude, density, syllable length, pitch, breath, noise, whisper, azimuth, and elevation over the render.

Fata Morgana Resynth

Use this for hybrid resynthesis from multiple input files. It analyzes selected items and recombines their timing, pitch, amplitude, and spatial traits into a new oscillator-field render.

Selection:

• Select 2-16 WAV-backed media items.
• Use sources with clearly different traits if you want the recombination to be obvious.

Important controls:

• Hybrid mode decides how sources are recombined.
• Trace behavior changes how partials are followed or redistributed.
• Traces per frame controls how much analysis detail is admitted.
• Trait mutation increases cross-source instability.
• Texture bias pushes the result toward denser or more textural behavior.
• Clarity protect helps avoid muddy low-frequency or overfull renders.

If the output becomes whistle-like, reduce pitch emphasis with fewer traces, lower pitch scale extremes, or stronger texture bias.

IR Toolkit

Use this to reshape a selected impulse response item before using it in a convolution process. Select one WAV-backed impulse item, then choose the edits to apply to the rendered copy.

Available operations:

• Trim silence removes leading or trailing low-level material.
• Tail fade applies an end fade over the selected fade length.
• Normalize sets the peak level of the processed impulse.
• Early reflections adds a sparse reflection pattern before the tail.
• Decorrelate channels gives multichannel IR channels slightly different timing and tone.

The output is a new media item, leaving the selected impulse item in place. Use this when an impulse needs a shorter tail, a defined peak level, or channel variation before convolution.

Karplus Field

Use this to render a multichannel field of plucked resonator events. The process does not require an input item. It creates short excitation impulses, feeds them through Karplus-Strong-style resonators, and distributes the resulting events across the selected channel count.

Main controls:

• Duration and Channels set the rendered item.
• Events and Density decide how many plucks are admitted.
• Pitch, Pitch spread, and Dispersion shape the resonator tuning.
• Damping and Brightness control decay and tone.
• Spatial width and motion controls distribute events across the channel bed.

Breakpoint envelopes can change density, pitch region, damping, and amplitude over the render. Use the amplitude envelope to make the field enter, thin out, or fade without changing the post-render item gain.

Mass Partial Field

Use this for additive fields made from many generated partial events rather than from a single analyzed source. Render time increases with duration, partial count, event count, and channel count.

Initial settings:

• Use a short duration while setting up a sound.
• Keep partial or event counts moderate.
• Leave normalize on.
• Shape amplitude and density with breakpoints before adding wide pitch drift.

This process is organized around mass behavior: many small related events that change density, register, and spatial focus over time. Render time increases with duration, partial count, event count, and channel count. The detailed breakpoint editor can be collapsed while setting other controls.

Modal Terrain

Use this for generated resonant fields made from many modal frequencies struck or excited over time. It is a source-free offline synth: the script creates the excitation events, resonant modes, and spatial distribution internally.

Output:

• 3OA ACN/SN3D writes a 16-channel third-order ambisonic item.
• Multichannel ring writes an even-channel ring bed.

Main controls:

• Mode count sets the size of the resonator bank.
• Events and Modes / event control how many excitations occur and how many modes each event activates.
• Frequency model selects inharmonic, metallic-clustered, stretched-harmonic, or scale-lattice tuning.
• Exciter changes the event shape from short strikes to dust, swells, or layered strata.
• Decay, Damping spread, Brightness, and Detune shape the modal tail and spectral motion.
• Spatial mode, Spatial width, and Motion determine how the resonances move through the chosen output format.

Breakpoint envelopes can shape amplitude, density, motion, and spatial width across the render. Use shorter durations and smaller mode counts while setting the behavior, then increase the mode and event counts for denser fields.

Partial Trace Resynth

Use this when you want an input file analyzed into sinusoidal traces and re-rendered as a multichannel oscillator field.

Selection:

• Select one WAV-backed media item.
• Run Partial Trace Resynth.

The process tracks peaks in the source spectrum, then resynthesizes selected traces with controllable density, trace behavior, pitch spread, and spatial motion. It is based on sine-wave resynthesis, so clear harmonic or resonant sources tend to reveal the method more directly than noisy sources.

Main controls:

• Trace mode changes whether partials remain linked, smear, point, or freeze.
• Density admits fewer or more traces before synthesis.
• Trace count or equivalent detail controls set how many peaks are followed.
• Pitch spread and jitter push the result away from literal resynthesis.
• Protection or clarity controls reduce buildup in dense renders.

If the output contains static-like edges, reduce density and pitch spread first. If it stays close to the source, increase trace motion or spatial spread.

Resonant Terrain

Use this for struck resonator banks, metallic fields, synthetic impulse responses, and sustained resonant dust. It sits between an offline synth and an impulse-design process.

Initial settings:

• Use moderate excitation.
• Use fewer resonators while learning the controls.
• Use amplitude and damping envelopes to shape the tail.
• Normalize to avoid unexpectedly quiet first renders.

Spread, detune, and channel motion determine how far the resonator bank moves away from a centered response. Excitation brightness and damping shape the upper-frequency edge before final gain is applied.

Subharmonic Bank

Use this to render a multichannel bank of divided oscillator voices. The process starts from one or more root tones, creates subharmonic divisions, and shapes them with masks, drift, fold, and pulse/sine balance.

Main controls:

• Duration and Channels set the rendered item.
• Root and Division range define the subharmonic material.
• Mask controls decide when divided voices are allowed through.
• Instability and Drift vary tuning and articulation over time.
• Pulse / sine blend changes the edge of the oscillator bank.
• Spatial spread distributes voices across the channel bed.

Breakpoint envelopes can change activity, amplitude, register, and spatial behavior over the render. Sparse mask settings create separated articulations; dense mask settings create more continuous divided-tone layers.