GlottDNN Vocoder

The GlottDNN package contains two main parts:

1) The glottal vocoder written in C++ - Dependencies: libsndfile, libgsl, libconfig

2) Python scripts for vocoder analysis, synthesis and training a DNN excitation model: - Dependencies: python, numpy, theano

Installing

The vocoder C++ code has the following library dependencies: - libgsl (GNU scientific library), for basic linear algebra and FFT etc. - libsndfile for reading and writing audio files - libconfig++ for reading structured configuration files

Usually the best way to install the dependencies is with the system package manager. For example, in Ubuntu 14.04, use apt-get install the packages libgsl0-dev, libsndfile1-dev, libconfig++-dev

The C++ part uses a standard GNU autotools build system. To compile the vocoder, run the following commands in this directory:

   ./configure
   make

Since the build targets are rather generically named Analysis and Synthesis, you might not want them in your default system PATH. To choose

   ./configure --prefix=/your/install/path/bin
   make install

Some typical use cases:

Train excitation model 16kHz sample rate

• modify config/config_default_16k.cfg
• modify python/config_default_16k.py
• run feature extraction and training script

python python/GlottDnnScript.py \
     python/config_default_16k.py

Please contact the authors for any questions, or open an issue at github:

Lauri Juvela: lauri.juvela@aalto.fi Manu Airaksinen: manu.airaksinen@aalto.fi

This code is licenced under the MIT licence, see LICENCE for more information