How to run locally 👇

1. go into generate (cd generate) and run python3 -m venv venv, then source venv/bin/activate, and then pip install -r requirements.txt
2. now we can start the flask server. to do this run python3 transcribe.py
3. set all environment variables in .env.example in .env with your own values.
4. we should now install the js deps with npm i.
5. inside generate/public/background, we must have the background video assets. At the bottom of this readme is the list of all the assets from a s3 bucket to download. Put all the videos in generate/public/background
6. now, run node localBuild.mjs and boom, a video locally generated in no time (actually in some time, 5-7 minutes typically)!

note: you need to create your own eleven labs voices and copy their voice id's. If you want to use Joe Rogan, Jordan Peterson, Barack Obama, and Ben Shapiro's voice you can go into generate/voice_training_audio to find the mp3 files to train your eleven labs voices with.

how to get google credentials:

this is probably the most complex api to get set up, so if you want to be able to generate videos with ai images instead of google fetched images, you only need open ai api credentials, and not google credentials

• https://developers.google.com/custom-search/v1/introduction/
• https://programmablesearchengine.google.com/controlpanel/all

how to get eleven labs credentials:

• https://elevenlabs.io/app/voice-library

how to get open ai credentials:

• https://platform.openai.com/api-keys

how to get groq api credentials:

• https://console.groq.com/keys

Video explaining how to run 👇 locally

assets to download

I have removed assets for download. If you want your own GTA / Minecraft / etc. bottom half video just find some on youtube.

common problems

• FFMPEG is not installed.
• You don't have the flask python server running (or not on port 5000)
• Dalle 3 API rate limit exceeded: this is because each dialogue transition has an image, and it is prompted to have 7 dialogue transitions. However, typical tier 1 open ai accounts can only generate 5 images per minute.
• You don't have folders public/srt and public/voice and src/tmp
• You have concurrency set too high for your computer (check remotion.config.ts)