Update app.py

app.py

@@ -1,5 +1,5 @@
 import subprocess
-import gradio as gr  # Add this import statement
+
 
 subprocess.run(["python", "-m", "pip", "install", "--upgrade", "pip"])
 subprocess.run(["pip", "install", "gradio", "--upgrade"])
@@ -8,50 +8,35 @@ subprocess.run(["pip", "install", "numpy"])
 subprocess.run(["pip", "install", "pydub"])
 subprocess.run(["pip", "install", "openai"])
 
-import gradio as gr
-import openai
-import soundfile as sf
-import numpy as np
-from pydub import AudioSegment
-from io import BytesIO
-
-# Set your OpenAI API key
-openai.api_key = "YOUR_OPENAI_API_KEY"
-
-# Whisper ASR model
-whisper_model = "whisper-small"
-
-# Define the Gradio interface
-iface = gr.Interface(
-    fn=None,  # To be defined later
-    inputs=gr.Audio(),
-    outputs=gr.Textbox(),
-    live=True,
-)
-
-# Define the function for ASR
-def transcribe_audio(audio_data):
-    # Convert the audio data to a suitable format
-    audio = AudioSegment.from_file(BytesIO(audio_data), format="wav")
-    audio.export("temp.wav", format="wav")
-    
-    # Load the audio file using soundfile
-    audio_array, _ = sf.read("temp.wav")
-
-    # Perform ASR using OpenAI's Whisper
-    response = openai.Completion.create(
-        engine=whisper_model,
-        audio_input=audio_array.tolist(),
-        content_type="audio/wav",
-    )
-
-    # Extract the transcribed text from the response
-    transcription = response["choices"][0]["text"].strip()
+import subprocess
 
-    return transcription  # This is the transcription result
+subprocess.run(["pip", "install", "datasets"])
+subprocess.run(["pip", "install", "transformers"])
+subprocess.run(["pip", "install", "torch", "torchvision", "torchaudio", "-f", "https://download.pytorch.org/whl/torch_stable.html"])
 
-# Set the function for the Gradio interface
-iface.fn = transcribe_audio
+import gradio as gr
+from transformers import WhisperProcessor, WhisperForConditionalGeneration
+
+# Load model and processor
+processor = WhisperProcessor.from_pretrained("openai/whisper-large")
+model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-large")
+model.config.forced_decoder_ids = None
+
+# Custom preprocessing function
+def preprocess_audio(audio_data):
+    # Apply any custom preprocessing to the audio data here if needed
+    return processor(audio_data, return_tensors="pt").input_features
+
+# Function to perform ASR on audio data
+def transcribe_audio(input_features):
+    # Generate token ids
+    predicted_ids = model.generate(input_features)
+    
+    # Decode token ids to text
+    transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
+    
+    return transcription[0]
 
-# Launch the Gradio app
-iface.launch()
+# Create Gradio interface
+audio_input = gr.Audio(preprocess=preprocess_audio)
+gr.Interface(fn=transcribe_audio, inputs=audio_input, outputs="text").launch()