Upload folder using huggingface_hub

.gitattributes

@@ -1,2 +1,7 @@
 assets/summary_plot.png filter=lfs diff=lfs merge=lfs -text
 model.weights.h5 filter=lfs diff=lfs merge=lfs -text
+X_test.npy filter=lfs diff=lfs merge=lfs -text
+X_train.npy filter=lfs diff=lfs merge=lfs -text
+best_emotion_model.keras filter=lfs diff=lfs merge=lfs -text
+emotion_recognition_model.keras filter=lfs diff=lfs merge=lfs -text
+emotion_recognition_wrapper_model.keras filter=lfs diff=lfs merge=lfs -text

X_test.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ecb935710198621df6f1bdb36dc4530670a1669ed9be3daebd865cb97ef326ab
+size 134535296

X_train.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3038748e3715c1a2e0f46ef61fe7e5207860c4404d7009ccace38d25fdf108f7
+size 537974912

best_emotion_model.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f7823e7ee344145eb360514117a4cc67e93c65b68547bc68a7f8c042404940b
+size 82992300

emotion_recognition_model.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58caaef2a85c3f86339dbbf5731110fdee9aaa05888b0b7319b4ea20dbc781e7
+size 82992300

emotion_recognition_wrapper_model.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b419cbe69b28b7f2630b6df7e5e924cb5b422d0da3d0debd9e73f89b40c23b08
+size 82983662

example_usage.py

@@ -0,0 +1,68 @@
+
+import tensorflow as tf
+import numpy as np
+import json
+import librosa
+import os
+
+def load_model(model_path):
+    "Load the emotion recognition model."
+    return tf.keras.models.load_model(model_path)
+
+def predict_emotion(model, audio_path, preprocessing_config):
+    "Predict emotion from an audio file."
+    # Load audio file
+    waveform, sr = librosa.load(
+        audio_path, 
+        sr=preprocessing_config["sample_rate"],
+        duration=preprocessing_config["duration"],
+        offset=preprocessing_config["offset"]
+    )
+    
+    # Ensure consistent length
+    target_length = int(preprocessing_config["sample_rate"] * preprocessing_config["duration"])
+    if len(waveform) < target_length:
+        waveform = np.pad(waveform, (0, target_length - len(waveform)))
+    if len(waveform) > target_length:
+        waveform = waveform[:target_length]
+    
+    # Extract features
+    mel_spec = librosa.feature.melspectrogram(
+        y=waveform,
+        sr=preprocessing_config["sample_rate"],
+        n_fft=preprocessing_config["frame_length"],
+        hop_length=preprocessing_config["hop_length"],
+        n_mels=128
+    )
+    
+    # Convert to log scale
+    log_mel = np.log(np.maximum(mel_spec, 1e-10))
+    
+    # Add batch and channel dimensions
+    features = np.expand_dims(np.expand_dims(log_mel, axis=0), axis=-1)
+    
+    # Make prediction
+    prediction = model.predict(features)[0]
+    
+    # Get emotion label
+    emotion_idx = np.argmax(prediction)
+    emotion = preprocessing_config["emotions"][str(emotion_idx)]
+    
+    return emotion, prediction[emotion_idx]
+
+# Example usage
+if __name__ == "__main__":
+    # Load model
+    model = load_model("emotion_recognition_model.keras")
+    
+    # Load preprocessing config
+    with open("preprocessing.json", "r") as f:
+        preprocessing_config = json.load(f)
+    
+    # Path to your audio file
+    audio_path = "path/to/your/audio.wav"
+    
+    # Predict emotion
+    emotion, confidence = predict_emotion(model, audio_path, preprocessing_config)
+    
+    print(f"Predicted emotion: {emotion} with confidence {confidence:.2f}")

preprocessing.json

@@ -0,0 +1,16 @@
+{
+  "sample_rate": 22050,
+  "duration": 2.5,
+  "offset": 0.6,
+  "frame_length": 2048,
+  "hop_length": 512,
+  "emotions": {
+    "0": "angry",
+    "1": "disgust",
+    "2": "fear",
+    "3": "happy",
+    "4": "neutral",
+    "5": "sad",
+    "6": "surprise"
+  }
+}