https://github.com/joaovitorlobo/youtube-videos-summarizer

YouTube Video Summarizer Powered by AI. Whisper-1 and GPT-4o-Mini implementation through OpenAI API. Audio download, transcription to text and summarizer.
https://github.com/joaovitorlobo/youtube-videos-summarizer

openai openai-api openai-gpt openai-whisper python python3 yt-dlp

Last synced: 3 months ago
JSON representation

YouTube Video Summarizer Powered by AI. Whisper-1 and GPT-4o-Mini implementation through OpenAI API. Audio download, transcription to text and summarizer.

• Host: GitHub
• URL: https://github.com/joaovitorlobo/youtube-videos-summarizer
• Owner: JoaoVitorLobo
• Created: 2025-02-26T13:19:20.000Z (3 months ago)
• Default Branch: main
• Last Pushed: 2025-02-26T13:38:07.000Z (3 months ago)
• Last Synced: 2025-02-26T14:31:32.749Z (3 months ago)
• Topics: openai, openai-api, openai-gpt, openai-whisper, python, python3, yt-dlp
• Homepage:
• Size: 2.93 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# YouTube-Videos-Summarizer 

# YouTube Video Summarizer Powered by AI. Whisper-1 and GPT-4o-Mini implementation through OpenAI API

from openai import OpenAI

from yt_dlp import YoutubeDL

import sys

# OpenAI API Key

key = "ENTER YOUR OPENAI API KEY HERE"

def download_wav(url: str, audio_file: str = "audio.wav") -> None:

    # Set URL download settings

    ydl_opts = {

        "format": "bestaudio/best",

        "outtmpl": audio_file,

        "overwrites": True # Overwrite audio_file if existing

    }

    

    # Download

    with YoutubeDL(ydl_opts) as ydl:

        ydl.download([url])

def speech_to_text(client: OpenAI, audio_file: str = "audio.wav") -> str:

    # Open audio_file

    with open(audio_file, "rb") as audio:

        # Transcribe using OpenAI's Whisper-1 model

        transcription = client.audio.transcriptions.create(model = "whisper-1", file = audio, response_format = "text")

    return transcription

def summarizer(transcription: str, client: OpenAI) -> str:

    # Summarize the transcription through chat with OpenAI's GPT-4o-mini model

    completion = client.chat.completions.create(

        model="gpt-4o-mini",

        messages=[

            {"role": "system", 

             # ChatGPT role

             "content": "You are a YouTube Videos Summarizer who will describe pragmatically a summary of the video's audio in its primary language."

             },

            {"role": "user",

             # Command to ChatGPT

             "content": f"Summarize this video transcript: {transcription}. Keep the summary concise, focusing only on key points and main takeaways."

            }   

            ])

    

    return completion

def main():

    # Download audio from URL

    download_wav(sys.argv[1])

    # Set OpenAI API client

    client = OpenAI(api_key= key)

    # Transcribe audio to text

    transcription = speech_to_text(client)

    # Summarize the transcription

    summary = summarizer(transcription, client)

    # Summary

    print(f"→ {summary.choices[0].message.content}")

main()