https://github.com/mrkrd/inverse_cochlea

Sound Decoding from Auditory Nerve Activity with Artificial Neural Networks
https://github.com/mrkrd/inverse_cochlea

artificial-neural-networks auditory cochlea mlp reconstruction sound spectrogram

Last synced: 10 months ago
JSON representation

Sound Decoding from Auditory Nerve Activity with Artificial Neural Networks

• Host: GitHub
• URL: https://github.com/mrkrd/inverse_cochlea
• Owner: mrkrd
• Created: 2019-07-25T10:33:44.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2019-07-25T10:47:42.000Z (over 6 years ago)
• Last Synced: 2025-03-29T11:41:40.644Z (11 months ago)
• Topics: artificial-neural-networks, auditory, cochlea, mlp, reconstruction, sound, spectrogram
• Language: Python
• Size: 1.13 MB
• Stars: 8
• Watchers: 1
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.rst

Awesome Lists containing this project

README

          
inverse_cochlea

===============

*inverse_cochlea* can reconstruct sounds from the activity of auditory

nerve fibers' using artificial neural networks::

  __|______|______|____    +-----------+

  _|________|______|___ -->|  Inverse  |      .-.     .-.     .-.

  ___|______|____|_____ -->|           |-->  /   \   /   \   /   \

  __|______|______|____ -->|  Cochlea  |          '-'     '-'

                           +-----------+

        ANF activity                                 Sound

Bernstein Conference 2012

-------------------------

Rudnicki M, Zuffo MK and Hemmert W (2012). Sound Decoding from

Auditory Nerve Activity. Front. Comput. Neurosci. Conference Abstract:

Bernstein Conference 2012. doi: 10.3389/conf.fncom.2012.55.00092

`Full abstract`_

`Sound samples`_

.. _`Full abstract`: https://www.frontiersin.org/10.3389/conf.fncom.2012.55.00092/event_abstract

.. _`Sound samples`: ./sound_samples

Poster

......

.. image:: ./poster/poster_2012.png

   :target: ./poster/poster_2012.pdf

Usage

-----

The direct reconstruction with an artificial neural network (suitable

for frequencies to up 2 kHz) is implemented in

inverse_cochlea.MlpReconstructor.  The reconstruction using a

combination of an artificial neural network and inverse spectrogram is

implemented in inverse_cochlea.ISgramReconstructor.

Both reconstructor classes can be configured by the constructor

parameters and have train() and run() methods.

To see how to use the package, have a look at the scripts in the

examples_ directory.

.. _examples: ./examples

Requirements

------------

- Python 2.7

- Numpy

- Scipy

- Pandas

- joblib

- cochlea_

For the  MlpReconstructor:

- ffnet

For the ISgramReconstructor:

- oct2py or pytave

- GNU Octave with ltfat_ preinstalled

.. _cochlea: https://github.com/mrkrd/cochlea

.. _ltfat: https://ltfat.github.io/

License

-------

The project is licensed under the GNU General Public License v3 or

later (GPLv3+).