Update app.py

app.py

@@ -1,10 +1,16 @@
 import gradio as gr
 import torch
+import torch.nn as nn
 import torch.nn.functional as F
 import torchaudio
 from transformers import AutoConfig, Wav2Vec2Processor, Wav2Vec2FeatureExtractor
 from src.models import Wav2Vec2ForSpeechClassification
+
+import librosa
+import IPython.display as ipd
 import numpy as np
+import pandas as pd
+import os
 
 model_name_or_path = "andromeda01111/Malayalam_SA"
 config = AutoConfig.from_pretrained(model_name_or_path)
@@ -12,67 +18,46 @@ feature_extractor = Wav2Vec2FeatureExtractor.from_pretrained(model_name_or_path)
 sampling_rate = feature_extractor.sampling_rate
 model = Wav2Vec2ForSpeechClassification.from_pretrained(model_name_or_path)
 
-# def speech_file_to_array_fn(path, sampling_rate):
-#     speech_array, _sampling_rate = torchaudio.load(path)
-#     resampler = torchaudio.transforms.Resample(_sampling_rate, sampling_rate)
-#     speech = resampler(speech_array).squeeze().numpy()
-#     return speech
-
-def speech_file_to_array_fn(audio_path):
-    if audio_path is None:
-        return None  # Handle cases where no file is provided
-    
-    try:
-        # Check if the input is a file path (upload) or direct audio data (recording)
-        if isinstance(audio_path, str):  
-            speech_array, _sampling_rate = torchaudio.load(audio_path)
-        else:  
-            # If it's recorded audio, Gradio provides it as a NumPy array
-            speech_array = torch.tensor(audio_path)  
-            _sampling_rate = sampling_rate  # Use default sampling rate
-        
-        # Resample to match model requirements
-        resampler = torchaudio.transforms.Resample(orig_freq=_sampling_rate, new_freq=sampling_rate)
-        speech = resampler(speech_array).squeeze().numpy()
-        return speech
-    
-    except Exception as e:
-        print(f"Error processing audio: {e}")
-        return None
-
-
-def predict(audio_path):
-    speech = speech_file_to_array_fn(audio_path, sampling_rate)
+
+def speech_file_to_array_fn(path, sampling_rate):
+    speech_array, _sampling_rate = torchaudio.load(path)
+    resampler = torchaudio.transforms.Resample(_sampling_rate)
+    speech = resampler(speech_array).squeeze().numpy()
+    return speech
+
+
+def predict(path, sampling_rate):
+    speech = speech_file_to_array_fn(path, sampling_rate)
     features = feature_extractor(speech, sampling_rate=sampling_rate, return_tensors="pt", padding=True)
-    
+
     input_values = features.input_values
     attention_mask = features.attention_mask
-    
+
     with torch.no_grad():
         logits = model(input_values, attention_mask=attention_mask).logits
-    
+
     scores = F.softmax(logits, dim=1).detach().cpu().numpy()[0]
-    output_emotion = {config.id2label[i]: f"{round(score * 100, 3):.1f}%" for i, score in enumerate(scores)}
-    
+    output_emotion = [{"Emotion": config.id2label[i], "Score": f"{round(score * 100, 3):.1f}%"} for i, score in enumerate(scores)]
+
     return output_emotion
 
+
 # Wrapper function for Gradio
 def gradio_predict(audio):
     predictions = predict(audio)
     return [f"{pred['Emotion']}: {pred['Score']}" for pred in predictions]
 
+
 # Gradio interface
 emotions = [config.id2label[i] for i in range(len(config.id2label))]
 outputs = [gr.Textbox(label=emotion, interactive=False) for emotion in emotions]
 
-# Gradio Interface with Audio Recording (max duration: 10 seconds)
 interface = gr.Interface(
     fn=predict,
-    inputs=gr.Audio(type="filepath", label="Record or Upload Audio"),
+    inputs=gr.Audio(label="Upload Audio", type="filepath"),
     outputs=outputs,
     title="Emotion Recognition",
-    description="Record or upload an audio file (max 10 sec) to predict emotions and their corresponding percentages.",
-    live=False,
+    description="Upload an audio file to predict emotions and their corresponding percentages.",
 )
 
 # Launch the app