Auxide

Auxide is a high-performance Rust CLI tool for image sonification, inspired by the encoding method used on the Voyager Golden Records. It converts images into audio waveforms by mapping pixel data to frequencies and can decode that audio back into the original images.

It leverages parallel processing (via rayon) to handle large batches of images efficiently.

Features

• Encode Images to Audio: Converts pixel intensity to audio frequency.
• Decode Audio to Images: Reconstructs images from WAV files using zero-crossing frequency analysis.
• Color Support: Supports both Grayscale and RGB (interleaved channels) modes.
• Batch Processing: Processes directories of files in parallel.
• Combination: Optionally combines multiple encoded images into a single audio track with spacers.

Installation

Ensure you have Rust and Cargo installed.

cargo build --release

The binary will be located at ./target/release/auxide.

Usage

Encoding (Image -> Audio)

By default, Auxide runs in encoding mode. It resizes images to the specified dimensions and generates a WAV file.

Basic Grayscale Encoding:

auxide --input ./images --output ./audio

Color Encoding: To encode Red, Green, and Blue channels sequentially:

auxide --input ./images --output ./audio --color

Combine into Single File: To merge all processed images into one long audio file (useful for transmission or creative effects):

auxide --input ./images --output ./audio --combine --spacer 2.0

Decoding (Audio -> Image)

To convert the audio back into an image, use the --decode flag.

Note: You must provide the same --width, --height, --duration-per-pixel, and --color settings used during encoding to reconstruct the image correctly.

auxide --decode --input ./audio --output ./reconstructed --width 512 --height 512

Options

Flag	Description	Default
-i, --input	Input directory or file path.	.
-o, --output	Output directory path.	./output
--width	Target width for resizing/reconstruction.	512
--height	Target height for resizing/reconstruction.	512
--color	Enable RGB color mode. If omitted, uses Grayscale.	false
--duration-per-pixel	Duration of the tone for each pixel (in seconds).	0.01
--min-frequency	Minimum frequency for pixel mapping (Hz).	200.0
--max-frequency	Maximum frequency for pixel mapping (Hz).	20000.0
--combine	Combine all encoded outputs into a single WAV file.	false
--spacer	Silence duration between combined images (in seconds).	1.0
--decode	Switch to decode mode (Audio -> Image).	false

How it Works

1. Encoding:

   • The image is resized to the target dimensions.
   • Pixels are read (either as Luma or RGB).
   • Pixel value (0-255) is mapped to a frequency range (default 200Hz - 20000Hz).
   • A sine wave is generated for the duration specified by --duration-per-pixel.
   • For RGB, the channels are encoded sequentially: Red, then Green, then Blue.

2. Decoding:

   • The audio is read in chunks corresponding to the pixel duration.
   • The frequency of each chunk is estimated using zero-crossing analysis.
   • The frequency is mapped back to a pixel value (0-255).
   • Pixels are assembled into the final image.