feat: Add support for multiple Whisper models

This commit adds a dropdown to select between different Whisper models for transcription. It includes the following changes:

- Add a dropdown input to the Gradio interface to select the model
- Pass the selected model ID to the transcribe_audio function
- Load the selected model and processor in the transcribe_audio function
- Update the pipeline to use the loaded model and processor

This allows users to easily switch between different Whisper models (e.g. large-v2, small-cantonese) for transcription based on their needs.

app.py

@@ -1,11 +1,11 @@
-# import spaces
+import spaces
 import gradio as gr
 # Use a pipeline as a high-level helper
 import torch
 from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
 # from datasets import load_dataset
 
-# @spaces.GPU(duration=120)
+@spaces.GPU(duration=120)
 def transcribe_audio(audio, model_id):
     if audio is None:
         return "Please upload an audio file."