About Teleprompter
Tele Prompter is a small open source project on GitHub that works as an on device teleprompter for conversation. Rather than scrolling a fixed script, it listens to what the user is saying and suggests quotes they can say next, with the stated aim of helping the speaker become more charismatic. The repository includes a demo video.
It is aimed at developers and hobbyists comfortable running Python code. By default it uses semantic embeddings to choose quotes, and the author also provides a highly experimental fine tuned language model that can be enabled by running the script with an llm argument. It depends on transformers and torch, installed through a standard conda setup, and is started by running main.py.
The project currently works only on macOS because it relies on the on device speech recognition of the hear library. The author notes that adding Whisper would be straightforward and invites pull requests. The repository is published under the MIT license by Daniel Gross.
There is no hosted service, and there is no cost to use it beyond running it locally.