tavox

tavox is a tool for adding (TTS-based) voice-over tracks to your LaTeX beamer presentations and for automatically generating videos out of them. It consists of two parts: a LaTeX package and a Python package.

Showcase

The LaTeX package introduces a few commands, one of which is speak. This command allows specifying the text to be said in the voice-over track for each individual slide of your presentation. Consider the example demo.tex shown below.

\documentclass{beamer}
\usepackage{tavox}

\begin{document}

\begin{frame}
	\speak{Hello! You are viewing an example presentation that shows how tavox is used.}
	Hello World!
\end{frame}

\end{document}

This example is deliberately kept simple to highlight the ease with which tavox can be integrated into your presentations. However, the speak command is actually much more powerful and also supports Beamer-style overlay specifications and optional arguments to introduce delays before and after the spoken lines. Please refer to the examples directory for more advanced demonstrations of the capabilities of tavox.

Upon compilation of the above presentation, a file named demo.tavox will be created by the package. This file can then be passed on to the accompanying tavox CLI tool that comes with the tavox Python package to create a video:

tavox demo.tavox

This command automatically synthesizes the required voice samples, renders the PDF file and creates an MLT project, which is then rendered into a video file. The created MLT project can also be opened in video editors like Shotcut or Kdenlive.

Installation

Run make install to install tavox Python package using pip and to copy the tavox LaTeX package to ~/texmf/tex/latex. Note that these two steps can also be run individually via make install_pip and make install_latex. In case you don't want the tavox latex package installed in your home directory, you can put it at an arbitrary location as long as you ensure that tavox.sty is in your LaTeX search path.

Dependencies

Strict Dependencies:

• ffmpeg
• Imagemagick (PDF to image conversion)
• MLT Multimedia Framework (used by many video editors, such as Shotcut or Kdenlive)

Optional TTS-Services:

• Coqui TTS (open source TTS Python tool)
• OpenAI's Python API (Note: using OpenAI's TTS service requires paid API access)

Note that you need at least one working TTS service to run tavox. However, you can also create new ones yourself.

Optional:

• Shotcut, Kdenlive (allows to view the generated mlt files)
• rubber (to build the examples)

Below you will find instructions for installing the dependencies on Fedora (40+) and Ubuntu (tested on 22.04).

Fedora

sudo dnf install ffmpeg ImageMagick mlt
sudo dnf install shotcut rubber # optional
pip install openai # optional

Note that Coqui TTS Version 0.22.0 requires a specific Python version (3.9):

sudo dnf install python3.9
python3.9 -m ensurepip
python3.9 -m pip install TTS

Ubuntu

sudo apt-get install ImageMagick melt
sudo apt-get install shotcut rubber # optional
pip install openai # optional

Be aware that this just installs the melt CLI tool from MLT and it appears that there does not exist an Ubuntu package for the whole framework. However, as of writing, a full installation of the MLT framework is not required, the shotcut package comes with its own version and Tavox only uses melt.

According to its installation guide Coqui TTS works on Ubuntu (18.04) with Python versions >=3.9,<3.12. On Ubuntu 22.04 (Jammy) Coqui TTS worked out of the box with Python 3.11.4 and 3.11.5. It can be installed via

pip3 install TTS

Converting PDFs to PNGs using ImageMagick in Ubuntu might produce an error. An appropriate fix can be found here.

Getting Started

To get started go through the examples presentations in the examples folder. The examples support the following Makefile targets:

• slides: creates the slides PDF
• transcript creates the transcript PDF (the file transcript_setup.tex configures the style of transcript)
• video: renders the video
• video_VOICE: renders the video using the specified VOICE (e.g. make video_alloy for the default OpenAI voice).

Licensing Information

Python Package

The tavox Python package is licensed under the Lesser General Public License (LGPL) v3.0.

For full licensing terms, see the included LICENSE-LGPL.txt or visit: https://www.gnu.org/licenses/lgpl-3.0.html

LaTeX Package

The tavox LaTeX package is licensed under the LaTeX Project Public License (LPPL) v1.3c.

For full licensing terms, see the included LICENSE-LPPL.txt or visit: https://www.latex-project.org/lppl/

Additional Notes

** Of course, all content you produce using tavox is your own! **