app.py

import os
from reportlab.lib.pagesizes import letter
from reportlab.pdfgen import canvas
import gradio as gr
import whisper

def transcribe_audio(audio_path: str, model_name: str):
    """
    Transcribes an audio file to text using the Whisper model and saves it as a PDF.

    Args:
        audio_path: The path to the audio file.
        model_name: The name of the Whisper model to use ("Cheetah", "Whale",
                    or "Dolphin").

    Returns:
        A tuple containing the transcribed text and the path to the generated PDF.
    """
    # Map the custom model names to actual Whisper model names
    model_map = {
        "Cheetah": "base",
        "Whale": "small",
        "Dolphin": "medium"
    }

    whisper_model_name = model_map.get(model_name, "base")  # Default to base if name not found

    # Load the Whisper model
    model = whisper.load_model(whisper_model_name)

    # Transcribe the audio
    result = model.transcribe(audio_path)
    transcribed_text = result["text"]

    # Create a PDF document
    pdf_filename = "transcription.pdf"
    c = canvas.Canvas(pdf_filename, pagesize=letter)

    # Add the transcribed text to the PDF
    textobject = c.beginText(50, 750)
    textobject.setFont("Helvetica", 12)
    # Split text into lines to fit on the page
    lines = transcribed_text.split('\n')
    for line in lines:
        textobject.textLine(line)

    c.drawText(textobject)

    # Save the PDF
    c.save()

    return transcribed_text, pdf_filename

# Define the Gradio interface
interface = gr.Interface(
    fn=transcribe_audio,
    inputs=[
        gr.Audio(type="filepath", label="Upload Audio File"),
        gr.Radio(["Cheetah", "Whale", "Dolphin"], label="Select Whisper Model", value="Cheetah")
    ],
    outputs=[
        gr.Textbox(label="Transcription"),
        gr.File(label="Download PDF")
    ],
    title="Audio Transcription App",
    description="Upload an audio file and select a Whisper model to get the transcription and a downloadable PDF."
)

# Launch the interface - Note: For deployment on Hugging Face Spaces,
# the interface will be launched automatically by the Space environment.
# You might not need the .launch() call in the final app.py for Spaces.
# However, for testing locally before deployment, keep it.
# For this app.py file intended for `gradio deploy`, we will remove the .launch()
# as the Space will handle the launch.

# interface.launch() # Commenting out .launch() for Spaces deployment