This Python script, video_subtitle_transformer.py, is designed to extract audio from video files, transcribe and translate the audio into a specified language using OpenAI's Whisper API, and then generate and embed timed subtitles directly into the video.

video_subtitle_transformer.py

import os
import moviepy.editor as mp
import subprocess
from openai import OpenAI
from pydub import AudioSegment

# Set OpenAI API Key
key = os.environ.get("OPENAI_API_KEY")
if not key:
    raise ValueError("OPENAI_API_KEY environment variable not set")
client = OpenAI(api_key=key)

def extract_audio_from_video(input_video, output_audio):
    video = mp.VideoFileClip(input_video)
    video.audio.write_audiofile(output_audio)
    print(f"Audio extracted: {os.path.exists(output_audio)}")
    return video.duration

def transcribe_and_translate_audio(audio_file, src_lang="hi", target_lang="en"):
    MAX_CHUNK_SIZE = 24 * 1024 * 1024  # 24 MB
    audio = AudioSegment.from_wav(audio_file)
    duration_ms = len(audio)
    chunk_duration_ms = (MAX_CHUNK_SIZE / len(audio.raw_data)) * duration_ms

    translated_text = ""
    for i, chunk_start in enumerate(range(0, duration_ms, int(chunk_duration_ms))):
        chunk_end = min(chunk_start + int(chunk_duration_ms), duration_ms)
        chunk = audio[chunk_start:chunk_end]
        
        chunk_file = f"temp_chunk_{i}.wav"
        chunk.export(chunk_file, format="wav")
        
        print(f"Processing chunk {i+1}, size: {os.path.getsize(chunk_file)} bytes")
        
        with open(chunk_file, "rb") as audio_chunk:
            response = client.audio.translations.create(
                model="whisper-1",
                file=audio_chunk,
                response_format="text"
            )
        translated_text += response + " "
        
        os.remove(chunk_file)

    return translated_text.strip()

def create_timed_subtitles(translated_text, video_duration):
    words = translated_text.split()
    total_words = len(words)
    words_per_second = total_words / video_duration
    
    subtitles = []
    current_subtitle = ""
    word_count = 0
    start_time = 0
    
    for word in words:
        current_subtitle += word + " "
        word_count += 1
        
        if word_count >= words_per_second * 3 or word.endswith('.'):
            end_time = start_time + (word_count / words_per_second)
            subtitles.append((start_time, end_time, current_subtitle.strip()))
            current_subtitle = ""
            word_count = 0
            start_time = end_time
    
    return subtitles

def format_time(seconds):
    hours = int(seconds // 3600)
    minutes = int((seconds % 3600) // 60)
    seconds = seconds % 60
    return f"{hours:02d}:{minutes:02d}:{seconds:06.3f}"

def create_srt_content(subtitles):
    srt_content = ""
    for i, (start, end, text) in enumerate(subtitles, 1):
        srt_content += f"{i}\n{format_time(start)} --> {format_time(end)}\n{text}\n\n"
    return srt_content

def add_subtitles_to_video(input_video, srt_content, output_video):
    srt_file = "temp_subtitles.srt"
    with open(srt_file, "w", encoding="utf-8") as f:
        f.write(srt_content)
    
    command = [
        "ffmpeg",
        "-i", input_video,
        "-vf", f"subtitles={srt_file}",
        "-c:a", "copy",
        output_video
    ]
    result = subprocess.run(command, capture_output=True, text=True)
    
    if result.returncode != 0:
        print(f"FFmpeg Error: {result.stderr}")
    else:
        print("Subtitles added successfully.")
        print(f"Output video created: {os.path.exists(output_video)}")
    
    os.remove(srt_file)

def process_video(input_video, output_video, src_lang="hi", target_lang="en"):
    audio_file = "temp_audio.wav"

    print(f"Extracting audio from video: {input_video}")
    video_duration = extract_audio_from_video(input_video, audio_file)

    print("Transcribing and translating audio...")
    translated_text = transcribe_and_translate_audio(audio_file, src_lang, target_lang)

    print("Creating timed subtitles...")
    timed_subtitles = create_timed_subtitles(translated_text, video_duration)

    print("Creating SRT content...")
    srt_content = create_srt_content(timed_subtitles)

    print("Adding subtitles to video...")
    add_subtitles_to_video(input_video, srt_content, output_video)

    # Clean up temporary files
    os.remove(audio_file)

    if os.path.exists(output_video):
        print(f"Process completed! Output saved to {output_video}")
    else:
        print(f"Error: Output file not created at {output_video}")

if __name__ == "__main__":
    input_video = "fieldmap_hindi.mp4"
    output_video = "fieldmap_hindi_subtitled.mp4"
    process_video(input_video, output_video, src_lang="hi", target_lang="en")