Introduction
Tele-Prompter is an on-device teleprompter that uses semantic retrieval together with local speech recognition to provide cue lines and prompts that help users improve spontaneity and delivery during presentations or recordings.
Key Features
- Local-first execution with dependencies on transformers and torch; demonstrated primarily on macOS.
- Semantic embedding-based retrieval for varied quote and prompt suggestions.
- Simple CLI entry point (
python main.py) and an included demo video illustrating usage.
Use Cases
- Presentation rehearsal and speaking practice to reduce pauses and improve flow.
- Privacy-sensitive video creators who prefer local, offline prompting during recording.
- Prototypes combining on-device speech recognition with semantic retrieval for research or demos.
Technical Highlights
- Implemented in Python with core dependencies on transformers and torch, and integrates the
hearlibrary for on-device speech recognition on macOS. - Released under the MIT license; small codebase with CLI and template-based front-end for quick experimentation.