Explore the synergy of audio processing and artistry in AudioFrequencyArtistry. Transform audio into visuals through stages, unlocking new dimensions of creativity. Copyright Melchor Lafuente Duque
This repository comprises an audio processing pipeline organized into three distinct stages: STAGE_1, STAGE_2, and STAGE_3. Each stage contains a set of scripts that transform audio signals into frequency domain representations using Discrete Cosine Transform (DCT), apply style transfer algorithms, reconstruct audio, and enhance audio quality through bandpass filtering.
Before you start working with the different stages, it's essential to download the necessary input files. Run the SETUP_INPUTS.ipynb notebook or SETUP_INPUTS.py script to download the required files.
The scripts within this stage focus on transforming audio signals into frequency domain representations using DCT. These representations are then utilized to reconstruct the audio signal and apply bandpass filters to enhance its quality.
1_Preprocessing_audio_to_image2_Processing_image_to_audio3_Postprocessing_bandpass_filter
The scripts within this stage focus on style transfer algorithms, audio reconstruction, and quality improvement. The goal is to create audio signals that resemble famous paintings through style transfer.
1_Preprocessing_audio_to_image2_Processing_style_transfer3_Processing_image_to_audio4_Postprocessing_bandpass_filter
In this stage, the scripts synchronize two audio files, apply style transfer with complex source and style images, and reconstruct audio with enhanced quality.
1_Preprocessing_modify_BPM2_Preprocessing_audio_to_image3_Processing_style_transfer4_Processing_image_to_audio5_Postprocessing_bandpass_filters