Cerebro

AI Powered (WhisperAI) - Simple CLI that allows you to search youtube video content given a search term and generates an exact Youtube URL with timestamp in the Youtube video wherever those terms are mentioned

Installation

Whisper-AI

This repository uses youtube-dl and OpenAI's Whisper to generate subtitle files for any youtube video.

To get started, you'll need Python 3.7 or newer. Install the binary by running the following command:

sudo -H pip install git+https://github.com/baseinfinity/yt-whisper.git

If after this you cannot run the yt_whisper globally from the CLI, then add the directory path that pip installs packages to your ~/.bash_profile or ~/.bashrc:

export PATH="~/.local/bin:$PATH"

Close your terminal to start a new session and account for these new changes.

ffmpeg

You'll also need to install ffmpeg, which is available from most package managers:

# on Ubuntu or Debian
sudo apt update && sudo apt install ffmpeg

# on MacOS using Homebrew (https://brew.sh/)
brew install ffmpeg

# on Windows using Chocolatey (https://chocolatey.org/)
choco install ffmpeg

Node

You will also need to install Node: >=14.0.0

Cerebro

Now that WhisperAI is installed, lets install other packages and generate a distro build

# If you haven't cloned the repo and aren't at root directory yet....
gh repo clone BaseInfinity/cerebro && cd cerebro

npm install && npm run build

Usage

./bin/run find "https://www.youtube.com/watch?v=dQw4w9WgXcQ" "up, give, you, gonna, never"

Example Output:

[youtube] dQw4w9WgXcQ: Downloading webpage
[youtube] dQw4w9WgXcQ: Downloading android player API JSON
[info] dQw4w9WgXcQ: Downloading 1 format(s): 251
[download] Destination: /var/folders/m4/wmggw2n90qq12py32x4790gh0000gn/T/dQw4w9WgXcQ.webm
[download] 100% of    3.28MiB in 00:00:00 at 8.13MiB/s
[ExtractAudio] Destination: /var/folders/m4/wmggw2n90qq12py32x4790gh0000gn/T/dQw4w9WgXcQ.mp3
Deleting original file /var/folders/m4/wmggw2n90qq12py32x4790gh0000gn/T/dQw4w9WgXcQ.webm (pass -k to keep)
Downloaded video "Rick Astley - Never Gonna Give You Up (Official Music Video)". Generating subtitles...
Detected language: English
100%|███████████████████████████████████████████████████████████████████████████████████████████████████████21204/21204 
[01:23<00:00, 252.80frames/s]
Saved VTT to /Users/stefanayala/cerebro/transcriptions/dQw4w9WgXcQ/Rick_Astley___Never_Gonna_Give_You_Up__Official_Music_Video.vtt
Found: "You"
https://www.youtube.com/watch?v=dQw4w9WgXcQ&t=22s

Found: "You"
https://www.youtube.com/watch?v=dQw4w9WgXcQ&t=30s

Found: "you"
https://www.youtube.com/watch?v=dQw4w9WgXcQ&t=34s

Found: "you"
https://www.youtube.com/watch?v=dQw4w9WgXcQ&t=39s

Found: "Never"
https://www.youtube.com/watch?v=dQw4w9WgXcQ&t=42s

Found: "gonna"
https://www.youtube.com/watch?v=dQw4w9WgXcQ&t=42s

Important

Be sure to use quotes around your CLI arguments
Initial downloads and transcriptions will be slow, but you can repeat the command again with diff search inputs and it will be blazingly fast

Shoutouts

This couldn't have been done without the awesome CLI made by https://github.com/m1guelpf/yt-whisper and OpenAI's underlying Whisper tech.

My contribution is just a glorified abuse of regex =)