app.py

import os
os.system("pip install git+https://github.com/openai/whisper.git")
import whisper

from pytube import YouTube
import gradio as gr

infer_model = whisper.load_model("tiny")


def infer(link: str, add_timestamps: bool) -> str:
    audio_path = download_audio(link)

    if not audio_path:
        return "Unable to process request."

    result = infer_model.transcribe(audio_path)

    title = "Content"
    try:
        title = audio_path.split("/")[-1]
        title = title.split(".")[0]
    except Exception as e:
        print(f"Unable to extract title. Exception {e}")

    if not add_timestamps:
        print(result["text"])
        return title + "\n" + result["text"]

    result_text = title + "\n"
    for segment in result["segments"]:
        result_text += f"{float(segment['start']):.2f}s - {float(segment['end']):.2f}s : {segment['text']}\n"
    return result_text.strip("\n")


def download_audio(link: str) -> str:
    try:
        yt = YouTube(link)
        stream = yt.streams.get_audio_only()
        audio_path = stream.download()
        print(audio_path)
        return audio_path
    except Exception as e:
        print(f"Unable to download file. Exception {e}")
        return ""


demo = gr.Interface(
    fn=infer,
    inputs=[gr.Textbox(label = "Youtube Link", placeholder="Copy link here"), gr.Checkbox(value=True, label="Add timestamps?")],
    outputs=[gr.Textbox(label = "Transcription", placeholder=" Expected processing time ~ Half the length of video. Hang tight!!")],
    examples=[ ["https://www.youtube.com/watch?v=KL2T0XRzWUI", False], ["https://www.youtube.com/watch?v=yGB_K_xlHdI", False], ["https://www.youtube.com/watch?v=dv9sgFHS2Do", True],]
)

demo.launch()