Skip to content

imilas/ddsp_loss_comparison

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

115 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Sound-Matching Experiments

Requirements

Please ensure the following libraries and tools are installed:

  • jax, flax, optax
  • dawdreamer
  • kiyamoto
  • jaxwt
  • scikit-image (skimage)
  • audax
  • sbx-rl — for reinforcement learning experiments
  • Qdax
  • transformers — from Hugging Face
  • py-dtw — for Dynamic Time Warping
  • PIX — for SSIM with JAX
  • streamlit — for running the survey interface
  • scikit-posthocs — for confidence interval diagrams
  • kaleido — for figure export

How to Run Experiments

Install Marimo to view and run the experiment notebooks interactively.


Important Notebooks

  • random_experiment.py
    Runs a random experiment with a given loss_fn, learning_rate, and program_id.
    Used by run_experiments.sh.

  • program_design.py
    Explore and tweak synthesizer programs.

  • loss_landscape_navigation.py
    Visualize and experiment with different loss functions.


Key Code Components

  • ./helper_funcs/
    Contains:

    • Loss function definitions
    • Plotting utilities
    • DSP helper functions
  • programs.json
    Contains all FAUST synthesizer programs.


Running the Hearing Tests

  1. Extract hearing_test.tar into the hearing_test/ directory.

  2. Run the survey via:

    python hearing_survey.py

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors