pdf2video.py is a Python script that combines

• (selected pages of) a PDF presentation, and
• a text script

into a video narrated by the Amazon Polly text-to-speech engine. It can be used to generate, for instance, educational videos.

Please see this sample video, produced with the tool, for a short introduction.

Using pdf2video.py requires the following external tools and services:

• Python version 3.6 or later.
• The pdfinfo and pdftoppm command line tools provided in the poppler PDF rendering library. In Ubuntu Linux, you can install these with sudo apt get poppler-utils.
• The ffmpeg command line tool from the FFmpeg framework. In Ubuntu Linux, you can install these with sudo apt get ffmpeg.
• Access to Amazon Web Services.
• The AWS Command Line Interface configured with a profile that can access the Polly service. To use the neural voices (recommended for the best quality), remember to select a region in which they are supported.

In the simplest case,

python3 pdf2video.py presentation.pdf script.txt video.mp4

converts the PDF file presentation.pdf and the script script.txt into the video video.mp4 narrated by the default voice (Amazon Polly standard voice Joanna in the current version). The video includes SRT subtitles that can be displayed by most video players. In addition, for HTML use, WebVTT subtitles are produced in a separate file as well.

The selected PDF pages as well as the narration voice can be changed easily. For instance, the sample video was produced witth the command

python3 pdf2video.py sample.pdf sample.txt --pages "1,2,4-6" --voice Matthew --neural --conversational sample.mp4

All the options can be printed with python3 pdf2video.py --help.

The script file is formatted as follows. The script for each presentation page starts with a line #page and the following text then contains the script. In the script text, one can use

• *text* to read text in an emphasized style,
• @xyz@ to spell xyz as characters,
• #high/text/ to use higher pitch for text,
• #low/text/ to use lower pitch for text,
• #n, where n is a positive integer, to have a pause of length of n*100ms,
• #ph/word/pronunciation/ spell the word with the X-SAMPA pronunciation, and
• #sub/text/subtitle/ to use subtitle as the subtitle instead of the spoken text,

Please see the file sample.txt file for examples.

The pdf2video.py tool is relased under the MIT License.