Welcome to Danial Kooshki’s Official Page

This is the official page showcasing my activities in Audio DSP, JUCE plugin development, and AI‑driven sound transformation.

Visit My GitHub Profile

Featured Projects

Selected works in DSP, JUCE, and DDSP/AI. Each card links to the GitHub repository.

Tri‑Beat — Polyrhythm/Polymeter with Shapes

JUCE • C++ • Audio DSP • Realtime

A performance‑oriented metronome & groove designer using concentric polygons to create polyrhythms and polymeters. Accents, one‑shot samples, and a global 4/4 clock for alignment.

DDSP Sax Transformer

PyTorch • DDSP • TorchScript/ONNX •

a Differentiable Digital Signal Processing (DDSP) Autoencoder that transforms the timbre of any input sound into a tenor saxophone. The model is trained on tenor saxophone recordings and their extracted fundamental frequency (f0) tracks.

IR Extractor — Deconvolution Toolkit

JUCE • FFT • Convolution • Tooling

A desktop utility to derive impulse responses (IR) from dry/wet recordings using spectral division. Includes trim/normalize and export to WAV for convolution reverbs.

Music Genre Classification

TensorFlow • Keras • CNN & MFCC • Audio ML

A machine learning pipeline for automatic music genre recognition. Combines MFCC feature-based dense networks and a VGG-style CNN on Mel-spectrograms. Trained on the GTZAN dataset and supports prediction from any uploaded audio file.

DDSP Timbre Transfer

PyTorch • DDSP • TorchScript/ONNX • Real-time

A general DDSP-based timbre transfer autoencoder that can morph the sound of any instrument or voice into another instrument in real time. This repository contains the 48kHz trained model, TorchScript/ONNX exports.

Shimmer Reverb

JUCE • C++ • ZITAREV • FAUST

A cinematic shimmer reverb using ZitaRev core + granular pitch-shift layers (octave, fifth, etc). This repository contains a minimal, Generic-GUI Audio Plugin (VST3 + Standalone). Tested on Windows 10/11 + Visual Studio 2022.