AudiImage

AudiImage is an Open Source Project which delivers a new dimension in acoustic analysis, especially when human hearing is to be mimicked and when machine hearing is to be performed. This repository is a modified version of the CARFAC project. Especially the SAI (Stabilized Auditory Image) implementation is in focus, the CARFAC (Cascade of Asymmetric Resonators with Fast-Acting Compression) itself is unchanged. For more information about Auditory Images have a look on our paper¹ for a brief overview or get in touch with Richard F. Lyon's textbook "Human and Machine Hearing: Extracting Meaning from Sound"²³, where chapter 21 gives you a comprehensive explanation and description.

Table of Contents

• Getting Started
  • Quick start
  • Requirements
• Contribution Guidelines
• About
  • Maintainers
  • CARFAC
  • Used Encryption
  • License

Getting Started

In the root folder, you can find the starter scripts, which serve as the entry point to AudiImage:

• RunComputationOfSai.m to easily run a computation of a SAI (get summary and data MAT files; mandatory to run first, if no SAI dataset is available).
• RunProcessingOfSai.m to run a (re-)processing of an existing SAI dataset (get movie, "grams", ...).
• ShowSai.mlapp is an interactive MATLAB App to analyze an existing SAI dataset.
• For the above listed scripts there are associated parameter files, which are mandatory to configure before running the scripts. The files are called ParametersToComputeSai.m, ParametersToProcessSai.m and ParametersToShowSai.m.

In script and helper folder you can find core and helper functions of AudiImage to run the computation and processing of SAI datasets.

In matlab folder you can find the MATLAB implementation of the CARFAC model and the modified SAI implementation.

The resources folder primarily contains image files used to enhance the appearance of markdown files.

The docs folder includes more detailed documentation on how to use AudiImage.

Note: To avoid any start-up difficulties adjust the PATH_TO_FFmpeg in matlab/MakeMovieFromPngsAndWav.m to match your specific installation.

Quick start

Requirements

• MATLAB without any toolboxes
• FFmpeg⁴
• optional⁵: MATLAB with Signal Processing Toolbox for spectrogram function

Contribution Guidelines

Please read our contribution guidelines.

About

Maintainers

• Michael Kuka (michael.kuka@de.bosch.com)

CARFAC

This section is taken from the original README.md with slight modifications to fit to this fork.

The CAR-FAC (cascade of asymmetric resonators with fast-acting compression) is a cochlear model implemented as an efficient sound processor, for mono, stereo, or multi-channel sound inputs.

This package includes the MATLAB implementations of the CARFAC model as well as code for computing Stabilized Auditory Image (SAI).

See the design doc for a more detailed discussion of the software design.

Used Encryption

This repository does not contain or use encryption algorithms.

License

AudiImage is open-sourced under the Apache-2.0 license. See the LICENSE file for details.

For a list of other open source components included in AudiImage, see the file 3rd-party-licenses.txt.

Footnotes

1. M. Kuka and M. Fischer, “Hearing Equivalent Signal Analysis by Auditory Images in Industrial Applications,” in Fortschritte der Akustik - DAGA 2024, Hannover, Mar. 2024, pp. 1218–1221. [Online]. Available: https://pub.dega-akustik.de/DAGA_2024/files/upload/paper/44.pdf ↩

2. R. F. Lyon, Human and Machine Hearing: Extracting Meaning from Sound. Cambridge: Cambridge University Press, 2017. doi: 10.1017/9781139051699. ↩

3. Richard F. Lyon's homepage, the textbook's homepage and find here the author‘s draft of the textbook. ↩

4. Adjust the PATH_TO_FFmpeg in matlab/MakeMovieFromPngsAndWav.m to match your specific installation. ↩

5. When you set the processing option processOptions.isProcessSpectrogram = false then you can run AudiImage with only the MATLAB Base License. ↩