Live Oral Translator on top of pdfpc

Presenter console with oral live translation (version: 1.1)

Authors or contributors: Stéphane Galland

1. What is pdfpc?

pdfpc (PDF Presenter Console) is a cross-platform presentation tool designed for speakers who need a professional edge. It leverages a multi-monitor setup to give you a private, real-time dashboard of slide previews, speaker notes, and a timer—all while your audience sees only the current slide.

• Dual‑screen presenter console: Your audience sees the current slide on the main screen; you see a comprehensive overview on your laptop, including an image of the next slide and the remaining presentation time.
• Integrated timer & notes: Display the current time, a countdown timer, and your speaking notes (plain or Markdown) to help you stay on track.
• Works with any PDF: Accepts PDFs created by nearly any presentation software (LaTeX/Beamer, PowerPoint, LibreOffice, etc.).
• Open source & cross‑platform: Available for GNU/Linux, macOS (via Homebrew or MacPorts), and Windows through WSL.

2. What is Pdfpc-live-translator?

pdfpc-live-translator is a companion tool for the pdfpc presentation software. It captures the presenter's voice in real time, performs speech‑to‑text recognition, displays the spoken text as a subtitle on the screen, and optionally translates the caption live into another language. The translated text appears as an overlay on the audience‑facing pdfpc window, making presentations accessible to multilingual audiences without requiring manual slide notes.

The default translation direction is English to Simplified Chinese, but the source and target languages can be reconfigured.

Major Features

🎤 Live Voice Capture & Speech‑to‑Text

• Uses the system microphone to capture the presenter's voice.
• Runs a local speech recognition engine to convert speech into text.
• Displays the recognized text (the subtitle) at the bottom of the presentation overlay.

🌐 Real‑time Translation

• Automatically translates the recognized text from a source language to a target language.
• Default configuration: English to Chinese (Simplified).
• Translations appear immediately below (or alongside) the original caption.
• Supports swapping source/target languages and adding other language pairs via configuration.

🖥️ pdfpc Overlay Integration

• Renders the captions and translations as a transparent overlay on top of the pdfpc presentation window (audience screen).
• Does not interfere with pdfpc's own presenter console or slide controls.
• Overlay position, font size, and background transparency are adjustable.

⌨️ Offline Mode

Uses on‑device speech recognition and machine translation models: no internet required.

🎛️ Simple Command‑Line Interface

Start the translator with default settings:

$> python3 start_talk_translate.py my-slides.pdf

3. Installation of the translation script

The following elements must be installed to have the live translation working.

3.1. System-wide installation

You must install Python and its virtual environment.

$> sudo apt install python3-full
$> sudo apt install python3-venv

3.2. Create Python virtual environment

Create the Python virtual environment to install the live translation libraries.

$> mkdir ~/bin/python3_environments
$> cd ~/bin/python3_environments
$> python3 -m venv live_translator

3.3. Install local libraries

Install the libraries in the virtual environment

$> ~/bin/python3_environments/live_translator/bin/pip install tk
$> ~/bin/python3_environments/live_translator/bin/pip install pydub
$> ~/bin/python3_environments/live_translator/bin/pip install screeninfo
$> ~/bin/python3_environments/live_translator/bin/pip install whisper
$> ~/bin/python3_environments/live_translator/bin/pip install faster-whisper

3.4. Install VOSK

VOSK is used for capturing the sound wave from the microphone ad converting offline it to text.

$> ~/bin/python3_environments/live_translator/bin/pip install vosk

• Download the English model from VOSK models and extract it.
• Update the launching script with the full path to the downloaded library.

3.5. Install PyAudio

PyAudio is used for reading the microphone.

$> sudo apt install portaudio19-dev
$> ~/bin/python3_environments/live_translator/bin/pip install pyaudio

3.6. Install ARGOS

ARGOS is used for translating offline the text from a source language to a target language.

$> ~/bin/python3_environments/live_translator/bin/pip install argostranslate

3.7. Install the English-Chinese package:

Below, you could specify the translation dictionary to be installed.

$> ~/bin/python3_environments/live_translator/bin/argospm install translate-en_zh

4. Run pdfpc-live-translator

4.1. Synopsis

start_talk_translate [options] <pdf-file>

4.2. Options

pdfpctool accepts the following options from the command line: