from transformers import pipeline import gradio as gr from video_downloader import download_video, download_youtube_video from moviepy.editor import AudioFileClip from moviepy.video.io.ffmpeg_tools import ffmpeg_extract_subclip import os from pydub import AudioSegment import re pipe = pipeline("automatic-speech-recognition", model="Artanis1551/whisper_romanian") def process_video(date): # If the date is not in YYYY-MM-DD format, return an error message date = date.strip() date_pattern = re.compile(r"\b\d{4}\d{2}\d{2}\b") if not date_pattern.match(date): video_path = download_youtube_video( "https://www.youtube.com/watch?v=dQw4w9WgXcQ" ) transcription = "Please enter a date in the format YYYY-MM-DD." return video_path, transcription try: # Download the video video_path = download_video(date) # Extract the first 30 seconds of the video short_video_path = f"short_{date}.mp4" ffmpeg_extract_subclip(video_path, 0, 30, targetname=short_video_path) video_path = short_video_path # Extract audio from the short video audio_path = f"audio_{date}.wav" AudioFileClip(short_video_path).write_audiofile(audio_path) audio = AudioSegment.from_wav(audio_path) with open(audio_path, "rb") as audio_file: audio = audio_file.read() transcription = pipe(audio)["text"] # Remove the audio file os.remove(audio_path) except: video_path = download_youtube_video( "https://www.youtube.com/watch?v=dQw4w9WgXcQ" ) transcription = "No decision was made on this date." return video_path, transcription iface = gr.Interface( fn=process_video, inputs=gr.inputs.Textbox(label="Date with format YYYYMMDD"), outputs=[ gr.outputs.Video(), gr.Textbox(lines=100, max_lines=100, interactive=True), ], title="Romanian Transcription Test", description="This app transcribes videos from the Romanian Parliament" + " on a given date. Only the first 30 seconds of the " + "video will be used if it is longer than that.\n\n" + "Example dates where there are videos available:" + " 20231122 and 20230522.", ) iface.launch()