Skip to content

Latest commit

 

History

History
54 lines (46 loc) · 4.04 KB

README.md

File metadata and controls

54 lines (46 loc) · 4.04 KB
Raw

 

PhaseEngine

An experimental FM-synth written in C# intended to be the successor to gdsFM, with lower CPU usage and more predictable resource allocation for use in game engines and demos. Leverages Godot Engine as its primary front-end.

Screenshots

Main View Wiring Grid and 6op PresetsrTables Bind Envelopes

Features and differences to gdsFM

New

  • Fixed-point phase accumulator (utilizing 12.20 precision by default)
  • Up to 8 operators per voice
  • More modular, delegate-based oscillator and operator functionality (Operators can behave like filters or DSP in addition to FM)
    • FM and Bitwise (AND/OR/XOR and Ring Modulation) operators
    • Traditional analog-style filters and wave folding operators
  • Envelopes with target audio levels for each phase, more like DX-style synths
  • Envelope hold phase after initial attack and before initial decay, similar to the initial delay phase in gdsFM. Stays constant under rate scaling.
  • All FM operators support Reface-style operator feedback.
  • More traditional noise generators, including an LFSR-based generator with selectable periodicity (create more buzzing effects) similar to 2a03, etc.
  • Variable length wavetables from 32 samples up to 1024 samples per bank. Selectable sample bank per-oscillator for FM and bitwise ops.
  • Parameter automation bind system, allowing instruments to adjust their parameters over time using a tracker envelope-like interface

Different

  • Based on more traditional paradigms, comparable to most other FM synths under the hood
    • Envelopes use attenuation in the log domain rather than mixing linear volume.
    • Only one envelope curve type supported (exponential).
    • Envelope state changes are specified in iteration rates (like most other FM synths) instead of pure length (like in gdsFM).
    • Modulating phase is done using simple addition, without piping the result through a "modulation technique" oscillator.
    • Algorithm processing order is done iteratively from the "top-down" rather than recursively from the carrier.
  • Fixed polyphony monotimbral chip layout, with operators allocated based on a chip/voice specification rather than dynamically per-note
  • Monotimbrality and fixed polyphony help simplify the design, allowing multiple instances to be specified and put on independent audio buses
  • Automatic or manual note prioritization system
  • Fixed internal clock rate of 48 KHz
  • Operator parameter specification is separated from its implementation, allowing more tracker-like control over individual notes (temporarily override params)
  • More traditional LFO; Reface-style rate table.
  • Variable-length microsample oscillator.
  • Automatic wiring of algorithms based on simpler connection descriptions, with algorithm presets from several classic FM synths.

To be implemented

  • Chip clocking to match various audio output rates, perhaps with optional sample interpolation
  • Pure, arbitrary length PCM Sample playback
  • "Linear" wavetable morphing operator
  • Possible second LFO as well as OPZ waveforms for tx81z support
  • Voice format specification (almost complete!)
  • rTable bind system to automate keyfollow/velocity changes to any parameter (including modifying existing bind envelopes)

May or may not be implemented or come back

  • Wider range of sampling options
  • Multitimbral operation (for now, multiple chip instances syncing their clocks should suffice)
  • High-quality oscillator lookup table support (greater than 10-bit), for richer timbres