rrossilli / youtext

A Streamlined YouTube Transcript Extractor API Service

YouText

Read succinct AI-interpreted summaries of dialogue-heavy youtube videos like podcasts, documentaries, tutorials, news, etc...

Why

Today, a vast portion of essential content and discourse is disseminated via online videos. While videos, including lectures and podcasts, offer an engaging medium, they often don't match the efficiency of reading. YouText aims to bridge this gap, providing users with a more streamlined method to absorb this information.

Tutorial

1. simply grab the youtube video id that you want to summarize

you can find this in the youtube video url:

                           here ↓↓↓↓↓↓↓↓↓↓↓
https://www.youtube.com/watch?v=L_Guz73e6fw&t=117s&ab_channel=LexFridman

or click the "share button"

            here ↓↓↓↓↓↓↓↓↓↓↓ 
https://youtu.be/L_Guz73e6fw

2. simply provide the video id after the youtext.io url

$ curl https://youtext.io/L_Guz73e6fw/interpretation

# specify view=1 to return the interpretation *only*
$ curl https://youtext.io/L_Guz73e6fw/interpretation?view=1

or for just the transcript

$ curl https://youtext.io/L_Guz73e6fw

WIP

1. summaries of dialogues (with more than 1 speaker) have formatting inconsistencies and may be difficult to read. there is plenty of room for improvement here: a. split summary into paragraphs for better readability b. TODO
2. probably should have written this in typescript :/
3. detect and omit ads

Contribution

Submit a PR!

Versioning

install bumpversion if you don't have it already

$ brew install bumpversion

incremement version using one of these (self explanatory)

$ make bumpversion-patch
$ make bumpversion-minor
$ make bumpversion-major

Development

run your own instance locally

Architecture

NodeJS, express, OpenAI API, Firebase

Setup

create .env and fill in values

$ cp .env.local .env

install deps

$ npm i

Run

run with npm script

$ npm start

Docker

run with docker

# build and run
$ docker build . -t rrossilli/youtext<version>
$ docker run -p 3000:3000 rrossilli/youtext<version>

# or with docker compose
$ docker-compose up --build

K8s

deploy to kubernetes cluster

$ kubectl apply -f k8s/deploy.<environment>.yml