Giter Site home page Giter Site logo

lcnn's Introduction

LCNN

A TensorFlow implementation of light convolutional neural network (LCNN or light CNN).
This implementation is aimed to classify between genuine speech (spoken by human) and playback speech (played by loud speaker). LCNN-based playback speech detection was firstly introduced by [1] and it has been served as a playback spoofing countermeasure to evaluate a "speech-enhancement-based" playback attack method [2].

Usage

Train

./main.py \
	  --train_genuine training_data_genuine_file_list.txt \   # all data should be saved as binary format with float type
	  --train_spoof training_data_spoof_file_list.txt \
	  --dev_genuine dev_data_genuine_file_list.txt \
	  --dev_spoof dev_data_spoof_file_list.txt \
	  --epoch  9 \
	  --batch_size 64 \
	  --lr 0.0001 \        # initial learning rate
	  --dlr 0.9 \          # if classification error rate increases, learning rate will be decreased by this rate
	  --keep_prob 0.5      # dropout



Each line of "*file_list.txt" contains a path to a corresponding data file. A data file contains a sequence of spectrogram (see [1] and [2] for details). "*genuine_file_list.txt" means file list of genuine speech and "*spoof_file_list.txt" means file list of playback speech. "dev_data*" means development dataset used for monitoring classification error rate.

A pre-trained model is included in "checkpoint" folder.

Test

./main.py \
	  --test_data test_data_file_list.txt \
	  --phase test \
	  --test_dir output_directory # all output is saved in binary format with float type

The output file contains a sequence of 2-D vectors and each vector is composed by two probabilities (genuine and playback).

Authors

Fuming Fang
National Institute of Informatics, Japan
[email protected]

License

BSD 3-Clause License

Reference

[1] Lavrentyeva, Galina, et al. "Audio replay attack detection with deep learning frameworks." Proc. Interspeech. 2017.
[2] F. Fang et al., "Transforming acoustic characteristics to deceive playback spoofing countermeasures of speaker verification systems," IEEE International Workshop on Information Forensics and Security (WIFS), 2018.

lcnn's People

Contributors

fangfm avatar

Watchers

 avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.