app.py

import gradio as gr
from pytube import YouTube
import subprocess
from huggingsound import SpeechRecognitionModel
import torch
import librosa
import soundfile as sf
from transformers import pipeline

def process_video(video_url):
    yt = YouTube(video_url)
    audio_file = yt.streams.filter(only_audio=True, file_extension='mp4').first().download(filename='ytaudio.mp4')
    subprocess.run(['ffmpeg', '-i', 'ytaudio.mp4', '-acodec', 'pcm_s16le', '-ar', '16000', 'ytaudio.wav'])
    
    device = "cuda" if torch.cuda.is_available() else "cpu"
    model = SpeechRecognitionModel("jonatasgrosman/wav2vec2-large-xlsr-53-english", device=device)
    
    input_file = 'ytaudio.wav'
    stream = librosa.stream(input_file, block_length=30, frame_length=16000, hop_length=16000)
    
    full_transcript = ''
    for i, speech in enumerate(stream):
        sf.write(f'{i}.wav', speech, 16000)
        transcription = model.transcribe([f'{i}.wav'])[0]['transcription']
        full_transcript += transcription + ' '

    summarization = pipeline('summarization')
    summarized_text = summarization(full_transcript, max_length=130, min_length=30, do_sample=False)
    return summarized_text[0]['summary_text']

iface = gr.Interface(fn=process_video, inputs="text", outputs="text", title="YouTube Video Summarizer")
iface.launch()