Update app.py

app.py

@@ -1,32 +1,28 @@
 import gradio as gr
 import torch
-from datasets import load_dataset, Audio
-from transformers import EncodecModel, AutoProcessor
-import spaces 
-from encodec import compress, decompress
+import torchaudio
+from transformers import EncodecModel
+from encodec.utils import convert_audio
+from encodec import compress_to_file, decompress_from_file
 import io
 
-# Load the Encodec model and processor
+# Load the Encodec model 
 model = EncodecModel.from_pretrained("facebook/encodec_48khz")
-processor = AutoProcessor.from_pretrained("facebook/encodec_48khz")
+model.set_target_bandwidth(6.0)  # Set the desired bandwidth (adjust as needed)
 
 @spaces.GPU
 def encode(audio_file_path):
     try:
-        # Open the audio file
-        with open(audio_file_path, "rb") as audio_file: 
-            # Load and preprocess the audio
-            audio_sample, sampling_rate = load_dataset("audiofolder", data_dir=audio_file_path, split="train")[0]["audio"]
-            wav = torch.tensor(audio_sample).unsqueeze(0)
+        # Load and pre-process the audio waveform
+        wav, sr = torchaudio.load(audio_file_path)
+        wav = convert_audio(wav, sr, model.sample_rate, model.channels)
 
-            # Compress to ecdc
-            compressed_audio = compress(model, wav)
+        # Compress to ecdc file in memory
+        output = io.BytesIO()
+        compress_to_file(model, wav, output) 
+        output.seek(0)
 
-            # Save compressed audio to BytesIO
-            output = io.BytesIO(compressed_audio)
-            output.seek(0)
-        
-            return output
+        return output
 
     except Exception as e:
         gr.Warning(f"An error occurred during encoding: {e}")
@@ -35,12 +31,9 @@ def encode(audio_file_path):
 @spaces.GPU
 def decode(compressed_audio_file):
     try:
-        # Load compressed audio
-        compressed_audio = compressed_audio_file.read()
-
         # Decompress audio
-        wav, sr = decompress(compressed_audio)
-        
+        wav, sr = decompress_from_file(compressed_audio_file)
+
         # Convert the decoded audio to a numpy array for Gradio output
         decoded_audio = wav.cpu().numpy()