app.py

from transformers import pipeline
import gradio as gr
import time
from video_downloader import download_video
from moviepy.editor import AudioFileClip
import datetime
import os

pipe = pipeline("automatic-speech-recognition", model="Artanis1551/whisper_romanian3")


def process_video(date):
    # Download the video
    video_path = download_video(date)

    # Extract audio from the video
    audio_path = f"audio_{date}.wav"
    AudioFileClip(video_path).write_audiofile(audio_path)

    # Transcribe the audio
    with open(audio_path, "rb") as audio_file:
        audio = audio_file.read()
    transcription = pipe(audio)["text"]

    # Remove the audio file
    os.remove(audio_path)

    return video_path, transcription


iface = gr.Interface(
    fn=process_video,
    inputs=gr.inputs.Textbox(label="Date with format YYYYMMDD"),
    outputs=[
        gr.outputs.Video(),
        gr.outputs.Textbox(lines=1000, max_lines=1000, interactive=True),
    ],
    title="Romanian Transcription Test",
)

iface.launch()