Prosolia

speech features extraction pipeline for prosody analysis

                                              +--->  binned      +
                                              |                  |
                                              +--->  energy      |
                                              |                  +---> .mat file
   wav  +---> filterbank +---> compression +------>  delta       |
        |                                     |                  |
        |                                     +--->  delta-delta |
        +---> pitch, probability of voicing                      +

• wav file as input, Matlab mat file as output

• gammatone filterbank (spectrogram-like from the Gammatone Filterbank Toolkit)

• compressed spectrogram (cubic root or log)

• delta, delta-delta and energy computed on spectrogram

• pitch estimation and probability of voicing (from Kaldi)

Usage

Once installed, prosolia is available as a command-line tool. Pipeline parameters are read from a configuration file. To get in, simply have a prosolia --help.

For exemple:

prosolia ./some_speech.wav -c ./prosolia.cfg -o some_features.mat

Installation

• On Windows, we recommend to install and use prosolia in a CygWin environment, because Kaldi is better supported there.

• Install Kaldi by following the instructions from here. Note that you don't need the whole Kaldi system but only the compute-kaldi-pitch-feats program.

  To compile only the required subpart of Kaldi (the featbin sources), you have to do (from the kaldi directory):

cd tools
./extras/check_dependencies.sh
make -j 4  # -j N do a parallel build on N CPUs
cd ../src
./configure
make depend -j 4
make featbin -j 4  # use "make -j 4" to compile the entire Kaldi

• Prosalia relies on the system library libsndfile. On Windows and OS X, it is installed automatically. On Linux, you need to install libsndfile using your distribution's package manager, for example sudo apt-get install libsndfile1.

• Using the Anaconda distribution of Python 3 (conda is also available independently of Anaconda with pip install conda; conda init):

conda install python=3.6 cffi numpy scipy matplotlib
python setup.py install

Licence

Copyright 2016, 2017 Mathieu Bernard

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program. If not, see http://www.gnu.org/licenses/.