mumodo

MUltiMOdal DOcument - import, analyze, plot and manage multimodal data

####################################################################
#                                                  __              #
#             ____ ___  __  ______ ___  ____  ____/ /___           #
#            / __ `__ \/ / / / __ `__ \/ __ \/ __  / __ \          #
#           / / / / / / /_/ / / / / / / /_/ / /_/ / /_/ /          #
#          /_/ /_/ /_/\__,_/_/ /_/ /_/\____/\__,_/\____/           #
#                                 Dialogue Systems Group           #
#                                 University of Bielefeld          #
#                                 www.dsg-bielefeld.de             #
####################################################################

Welcome to MUltiMOdal DOcuments (MUMODO), the multimodal data management, post-processing and analysis Python package. Mumodo is a piece of software actively developed by the DSG (Dialogue Systems Group) in order to meet ALL of (y)our multimodal analysis needs. Mumodo makes extensive use of the Pandas library (http://pandas.pydata.org/) and its indexed tables (Dataframes); the IPython interpreter, as well as its html-based version, notebook; and Matplotlib, which mumodo uses for plotting. Other weak dependencies include moviepy to handle audio and video files, as well as PIL (Python Image Library) to handle image data.

We recommend installing latest stable versions of IPython, Pandas, and Matplotlib in order to take advantage of all the newest features.

In addition, mumodo makes use of the textgrid tools (tgt) package (https://github.com/hbuschme/TextGridTools), in order to import/export Praat textgrids.

Since the running enviroment of mumodo is the Python interpreter, mumodo users have access to the wealth of libraries available for scientific computing in Python, while enjoying the flexibility of the Python programming language in their analysis workflows. In addition, the IPython-Matplotlib combo allows for inline plots and markup text to be input around the actual commands and expressions that consitute the analysis, essentially documenting the analysis "in place". This helps tremendously in sharing research work and making it reproducible.

---

Mumodo Notebooks

The notebooks provided need to be loaded into an active IPython notebook session which is executed from a shell (Terminal in Mac OS or command prompt in Windows) as follows:

ipython notebook

Demo notebooks distributed with mumodo can then be loaded from the IPython dashboard. These demo notebooks showcase the basic funcionality of the basic mumodo packages. Their titles are self-explanatory and they contain instructions within themselves.

Sample data

Some sample data is distributed with mumodo. This data is required in order to run the demo notebooks. We recommend to move this data away from the installation directory to a folder in your "workspace" (see below).

---

INSTALLATION

There are threeinstallation methods. The standard way is to use pip. Alternatively you can clone mumodo directly using git and either run a setup script, or copy files manually.

A. INSTALLATION WITH PIP

TODO: Register mumodo on PIP

B. INSTALLATION from git

1. Get mumodo by cloning this repository to your local HDD, to a location of your choice.

2. Run the following command (You may need admin(root) permissions)

python setup.py install --install-data={directory}

{directory} is where the sample data and notebooks will go (your workspace)

NOTE: If you do not define the install-data variable, the sample data and notebooks will be copied to an OS-dependent, default place, under a folder called "mumodo". As this is less than ideal, we recommend you setup a workspace directory at this time. Obvious choices are your home folder, Documents folder, or a subfolder thereof

C. MANUAL INSTALLATION

1. Same as B.1

2. Add the "packages/" directory to your PYTHONPATH, or copy the "mumodo" subfolder therein to your "site-packages" folder. A third option is to use IPython profiles to add mumodo to the path via the startup profile functionality (check IPython documentation)

3. Create a "workspace" directory on your local HDD. You can have several sub-folders therein for individual projects. Copy the demo notebooks and sample data into this workspace. The notebooks will work if the sampledata folder is in their directory.

4. Install the tgt package. Get it from https://github.com/hbuschme/TextGridTools, or simply do

   pip install tgt

   mumodo has been tested with version 1.02 of tgt

---

USING MUMODO IN IPYTHON

Open a terminal, console (command prompt in windows), and go (cd) to your workspace (sub) directory, where you copied the notebooks and data, and run:

ipython notebook

The ipython dashboard should appear in your default webbrowser. Click on one of the demo noteboos to open it. A new tab should appear in the browser, that contains the new notebook. Check IPython documentation in order to get the basics of using IPython notebooks.

NOTE: The demo notebooks are not python scripts. You should not attempt to run all the cells at once. Rather, run one cell at a time and see what results it produces.

NOTE: You should no longer use the "-pylab" and "inline" arguments when starting ipython notebook, but rather follow the new guideline of beginning each notebook with:

 %matplotlib inline
 import numpy as np
 import matplotilb.pylab as plt

UPDATE 25.04.2015 The notebooks have been tested with IPython 3.1, and matplotlib 1.4.3.3 with no problems All packages and notebooks tested against pandas 0.16.0-2