CITATION.cff

cff-version: 1.2.0
title: "Spectro: Spectral State-Guided Synthesis (SSGS)"
message: "If you use Spectro or the SSGS algorithm in your research or software, please cite it using this metadata."
type: software

authors:
  - family-names: Davison
    given-names: Damien
  - family-names: Maillet
    given-names: Michael
  - family-names: Davison
    given-names: Sacha

abstract: >
  Spectro implements the Spectral State-Guided Synthesis (SSGS) algorithm: a 
  hybrid generative audio system that decouples high-level temporal structure 
  from low-level spectral content using Hidden Markov Models (HMMs), 
  Linear Predictive Coding (LPC), Karplus–Strong excitation, and A* search. 
  The model learns spectral states from LPC coefficient clusters, optimizes 
  transitions through EM-based HMM refinement, and synthesizes new audio by 
  decoding optimal state sequences and filtering state-conditioned excitation 
  signals. This architecture enables controllable, interpretable, and 
  data-efficient procedural sound generation.

version: "1.0.0"
date-released: 2025-01-01
license: "Apache-2.0"

repository-code: "https://github.com/recursive-ai-dev/spectro"
url: "https://github.com/recursive-ai-dev/spectro"