Update app.py

app.py

@@ -1,3 +1,45 @@
 import gradio as gr
+from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+from openai.whisper import WhisperProcessor
 
-gr.Interface.load("models/openai/whisper-medium").launch()
+# Load the model and tokenizer
+model_id = "openai/whisper-medium"
+model = AutoModelForSeq2SeqLM.from_pretrained(model_id)
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+
+# Create a WhisperProcessor instance
+processor = WhisperProcessor(model=model, tokenizer=tokenizer)
+
+# Define a function that takes an audio input and returns a transcription
+def transcribe(audio):
+  # Use the processor to transcribe the audio
+  transcription = processor.transcribe(audio)
+    
+  # Extract the confidence score and the duration from the transcription
+  confidence = transcription.confidence
+  duration = transcription.duration
+    
+  # Remove the special tokens from the transcription text
+  text = transcription.text.replace("<|startoftranscript|>", "").replace("<|endoftranscript|>", "")
+    
+  # Return the text, confidence and duration as outputs
+  return text, confidence, duration
+
+# Create a Gradio interface with two modes: realtime and file upload
+iface = gr.Interface(
+  fn=transcribe,
+  inputs=[
+    gr.inputs.Audio(source="microphone", type="numpy", label="Realtime Mode"),
+    gr.inputs.Audio(source="upload", type="numpy", label="File Upload Mode")
+  ],
+  outputs=[
+    gr.outputs.Textbox(label="Transcription"),
+    gr.outputs.Textbox(label="Confidence Score"),
+    gr.outputs.Textbox(label="Duration (seconds)")
+  ],
+  title="Whisper Transcription App",
+  description="A Gradio app that uses OpenAI's whisper model to transcribe audio"
+)
+
+# Launch the app
+iface.launch()