Spaces:
Sleeping
Sleeping
File size: 3,484 Bytes
2e56c11 b13d50e 06bb06f b13d50e 06bb06f b13d50e 2e56c11 b13d50e fd2d49f 87d87dd 2e56c11 b13d50e dc26eac b13d50e 1012200 dc26eac 1012200 b13d50e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 |
from youtube_transcript_api import YouTubeTranscriptApi
from urllib.parse import urlparse, parse_qs
from pytube import YouTube
from huggingface_hub import InferenceClient
import gradio as gr
model_name = "mistralai/Mixtral-8x7B-Instruct-v0.1"
client = InferenceClient(model=model_name)
def transcribe_video(url):
video_id = parse_youtube_url(url)
if video_id:
video_metadata = get_video_metadata(video_id)
transcript_content = get_transcript_content(video_id)
transcript_summary = summarise_transcript(transcript_content)
return (
f"Title: {video_metadata['title']}\nAuthor: {video_metadata['author']}",
transcript_content,
transcript_summary,
)
else:
return None
def parse_youtube_url(url):
parsed_url = urlparse(url)
video_id = parse_qs(parsed_url.query).get("v")
if video_id:
return video_id[0]
return None
def get_video_metadata(video_id):
yt = YouTube(f"https://www.youtube.com/watch?v={video_id}")
title = yt.title or "Unknown"
author = yt.author or "Unknown"
metadata = {"title": title, "author": author}
return metadata
def get_transcript_content(video_id):
try:
transcript = YouTubeTranscriptApi.get_transcript(video_id)
transcript_content = parse_transcript(transcript)
return transcript_content
except Exception as e:
raise e
def parse_transcript(transcript):
content = " ".join(
map(
lambda transcript_piece: transcript_piece["text"].strip(" "),
transcript,
)
)
return content
def summarise_transcript(transcript_content):
prompt = f"""Provide a summary of the following video transcription in 150-350 words, focusing on the key points and core ideas discussed: {transcript_content}"""
message = [{"role": "user", "content": prompt}]
result = client.chat_completion(
messages=message,
max_tokens=2048,
temperature=0.1,
)
return result.choices[0].message["content"].strip()
with gr.Blocks(theme=gr.themes.Base()) as demo:
gr.Markdown("<H1>YoutTube Summariser</H1>")
gr.Markdown(
"<H3>Provide a link to a YouTube video and get a transcription and summary</H3>"
)
gr.Markdown(
"<H6>Update: Recently YouTube has been cracking down on requests and blacklisting known cloud providers. Unfortunately, it seems that requests to YouTube APIs from HuggingFace are currently blocked. If you would still like to try out the summariser, clone the repo and run it locally with gradio app.py</H6>"
)
with gr.Row():
with gr.Column(scale=1):
video_link = gr.Textbox(
label="Link to video",
value="https://www.youtube.com/watch?v=ZIyB9e_7a4c",
)
transcribe_btn = gr.Button(
value="Transcribe & Summarise ⚡️", variant="primary"
)
with gr.Column(scale=5):
video_info = gr.Textbox(label="Video Info")
transcription = gr.TextArea(
label="Transcription", scale=1, lines=12, max_lines=12
)
transcription_summary = gr.TextArea(
label="Summary", scale=1, lines=12, max_lines=12
)
transcribe_btn.click(
fn=transcribe_video,
inputs=video_link,
outputs=[video_info, transcription, transcription_summary],
)
demo.launch()
|