app.py

import gradio as gr
import whisper
import os
from pydub import AudioSegment

def convert_video_to_audio(video_file):
    audio_file = video_file.rsplit('.', 1)[0] + ".wav"
    audio = AudioSegment.from_file(video_file)
    audio.export(audio_file, format="wav")
    return audio_file
    
def chunk_audio(audio_file, chunk_size_mb=25, max_duration_seconds=900):
    audio = AudioSegment.from_file(audio_file)
    chunk_size_bytes = chunk_size_mb * 1024 * 1024
    chunk_duration_ms = min(len(audio), max_duration_seconds * 1000)

    if len(audio) > chunk_duration_ms or audio.frame_count() * audio.frame_width > chunk_size_bytes:
        return [audio[:chunk_duration_ms]]
    else:
        return [audio]
      
model = whisper.load_model("base")

def transcribe_audio(file):
    if file is None:
        return "Error: No file uploaded.", None

    file_extension = os.path.splitext(file.name)[1].lower()

    if file_extension in ['.mp4', '.avi', '.mov']:
        audio_file = convert_video_to_audio(file.name)
    else:
        audio_file = file.name

    chunks = chunk_audio(audio_file)
    transcriptions = []

    for i, chunk in enumerate(chunks):
        chunk_file = f"chunk_{i}.wav"
        chunk.export(chunk_file, format="wav")
        result = model.transcribe(chunk_file)
        transcriptions.append(result["text"])
        os.remove(chunk_file)

    if file_extension in ['.mp4', '.avi', '.mov']:
        os.remove(audio_file)

    full_transcription = " ".join(transcriptions)

    output_filename = os.path.splitext(os.path.basename(file.name))[0] + ".txt"
    with open(output_filename, "w") as text_file:
        text_file.write(full_transcription)

    return full_transcription, output_filename

# Test function
def test_transcription(file_path):
    class MockFile:
        def __init__(self, path):
            self.name = path

    mock_file = MockFile(file_path)
    transcription, output_file = transcribe_audio(mock_file)
    print("Transcription:")
    print(transcription)
    print("\nOutput file:", output_file)

# Example usage (uncomment and replace with your file path)
# test_transcription("/path/to/your/audio_or_video_file.mp3")

# Gradio interface 
iface = gr.Interface(
    fn=transcribe_audio,
    inputs=gr.File(label="Upload Audio/Video File"),
    outputs=[
        gr.Textbox(label="Transcription"),
        gr.File(label="Download Transcript")
    ],
    title="QuickTranscribe AI",
    description="Upload an audio or video file to get its transcription. The transcript will be displayed and available for download. Files will be processed in chunks of 25MB or 15 minutes maximum. Please use responsibly."
)

iface.launch(share=True)