Awesome

cascaded-am-tuning-for-sound-recognition

Codes for "Cascaded Tuning to Amplitude Modulation for Natural Sound Recognition" by Takuya Koumura, Hiroki Terashima, and Shigeto Furukawa.

Reference

Koumura T, Terashima H, Furukawa S (2019) Cascaded Tuning to Amplitude Modulation for Natural Sound Recognition. J Neurosci 39(28):5517–5533. DOI: https://doi.org/10.1523/JNEUROSCI.2914-18.2019

Dependencies

• Python 3
• Chainer https://chainer.org/
• soundfile https://pypi.org/project/SoundFile/

Datasets

• The directory tree should look like this:

  cascaded-am-tuning-for-sound-recognition
  ├── cascaded-am-tuning-for-sound-recognition
  │   ├── am-meta-analysis
  │   │   └── cumulative
  │   │       ├── Atencio2010
  │   │       │   ├── txt files
  │   │       │   └── ...
  │   │       ├── Bartlett2007
  │   │       └── (more directories...)
  │   ├── ESC50
  │   │   ├── ESC-50
  │   │   │   ├── 101 - Dog
  │   │       │   ├── ogg files
  │   │       │   └── ...
  │   │   │   ├── 102 - Rooster
  │   │   │   └── (more directories...)
  │   │   ├── info.txt
  │   │   └── Results
  │   │       ├── Result0
  │   │       │   ├── Architecture.txt
  │   │       │   ├── ConfusionMatrix.txt
  │   │       │   ├── Params.txt
  │   │       │   ├── Response
  │   │       │   │   ├── Am
  │   │       │   │   └── Am0
  │   │       │   └── TrainedModel
  │   │       ├── Result1
  │   │       │   └── (same filenames as in Result0)
  │   │       └── (more directories...)
  │   └── TIMIT
  │       ├── Results
  │       │   ├── Result0
  │       │   │   ├── Architecture.txt
  │       │   │   ├── ConfusionMatrix.txt
  │       │   │   ├── Params.txt
  │       │   │   ├── Response
  │       │   │   │   ├── Am
  │       │   │   │   └── Am0
  │       │   │   └── TrainedModel
  │       │   ├── Result1
  │       │   │   └── (same filenames as in Result0)
  │       │   └── (more directories...)
  │       └── TIMIT
  │           ├── TEST
  │           │   ├── DR1
  │           │   ├── DR2
  │           │   └── (more directories...)
  │           └── TRAIN
  │               ├── DR1
  │               └── (more directories...)
  ├── draw_mtf_esc50.py
  ├── draw_mtf_timit.py
  └── (more py files...)
  

  (It may be a bit confusing because directories with the same name appears twice...)

• ESC-50: Dataset for Environmental Sound Classification https://doi.org/10.7910/DVN/YDEPUT

  • Download audio files and put them in the directory "ESC50" indluced in the above dataset at figshare
  • Note: our code is built for the older version of ESC-50 with ogg format, in which folder organizations are slightly different from the current version at https://github.com/karoldvl/ESC-50

• TIMIT Acoustic-Phonetic Continuous Speech Corpus https://catalog.ldc.upenn.edu/LDC93S1

  • Download audio files and put them in the directory "TIMIT" indluced in the above dataset at figshare

License

Please see LICENSE.