Added the application file
Browse files
app.py
ADDED
@@ -0,0 +1,121 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import joblib
|
3 |
+
import numpy as np
|
4 |
+
import xgboost
|
5 |
+
# from sklearn.ensemble import GradientBoostingClassifier
|
6 |
+
# from tensorflow.keras.models import load_model
|
7 |
+
import tensorflow
|
8 |
+
from keras.losses import binary_crossentropy
|
9 |
+
from keras.optimizers import Adam
|
10 |
+
from tensorflow import keras
|
11 |
+
from keras.models import load_model
|
12 |
+
# Local Imports
|
13 |
+
import feature_extraction
|
14 |
+
import audio_splitting
|
15 |
+
|
16 |
+
# Create a Streamlit web app
|
17 |
+
st.title("Music Genre Classifier")
|
18 |
+
|
19 |
+
# Upload music file
|
20 |
+
uploaded_file = st.file_uploader("Upload a music file", type=["mp3", "wav"])
|
21 |
+
|
22 |
+
if uploaded_file is not None:
|
23 |
+
# User selects a model
|
24 |
+
all_models = ["K-Nearest Neighbors - (Single Label)", "Logistic Regression - (Single Label)", "Support Vector Machines - (Single Label)",
|
25 |
+
"Neural Network - (Single Label)",
|
26 |
+
"XGB Classifier - (Single Label)", "Convolutional Recurrent Neural Network - (Multi Label)", "XGB - (Multi Label)",
|
27 |
+
"Neural Network - (Multi Label)","Batch Normalization - (Multi Label)"]
|
28 |
+
model_name = st.selectbox("Select a model", all_models)
|
29 |
+
st.write(f"Predicition of following genres")
|
30 |
+
multi_class_names = ["Metal", "Jazz", "Blues", "R&B", "Classical", "Reggae", "Rap & Hip-Hop", "Punk", "Rock",
|
31 |
+
"Country", "Bebop", "Pop", "Soul", "Dance & Electronic", "Folk"]
|
32 |
+
|
33 |
+
st.write(multi_class_names)
|
34 |
+
|
35 |
+
# Load the selected model
|
36 |
+
if model_name == "K-Nearest Neighbors - (Single Label)":
|
37 |
+
model = joblib.load("./models/knn.pkl")
|
38 |
+
elif model_name == "Logistic Regression - (Single Label)":
|
39 |
+
model = joblib.load("./models/logistic.pkl")
|
40 |
+
elif model_name == "Support Vector Machines - (Single Label)":
|
41 |
+
model = joblib.load("./models/svm.pkl")
|
42 |
+
elif model_name == "Neural Network - (Single Label)":
|
43 |
+
model = joblib.load("./models/nn.pkl")
|
44 |
+
elif model_name == "XGB Classifier - (Single Label)":
|
45 |
+
model = joblib.load("./models/xgb.pkl")
|
46 |
+
elif model_name == "XGB - (Multi Label)":
|
47 |
+
model = joblib.load("./models/xgb_mlb.pkl")
|
48 |
+
elif model_name == "Convolutional Recurrent Neural Network - (Multi Label)":
|
49 |
+
model = tensorflow.keras.models.load_model("./models/model_crnn1.h5", compile=False)
|
50 |
+
model.compile(loss=binary_crossentropy,
|
51 |
+
optimizer=Adam(),
|
52 |
+
metrics=['accuracy'])
|
53 |
+
elif model_name == "Neural Network - (Multi Label)":
|
54 |
+
model = tensorflow.keras.models.load_model("./models/model_nn.h5", compile=False)
|
55 |
+
model.compile(loss=binary_crossentropy,
|
56 |
+
optimizer=Adam(),
|
57 |
+
metrics=['accuracy'])
|
58 |
+
elif model_name == "Batch Normalization - (Multi Label)":
|
59 |
+
model = tensorflow.keras.models.load_model("./models/model_bn.h5", compile=False)
|
60 |
+
model.compile(loss=binary_crossentropy,
|
61 |
+
optimizer=Adam(),
|
62 |
+
metrics=['accuracy'])
|
63 |
+
class_names = ["blues", "classical", "country", "disco", "hiphop", "jazz", "metal", "pop", "reggae", "rock"]
|
64 |
+
|
65 |
+
xgb_multi_class_names = ["Rock", "Rap & Hip-Hop", "Soul", "Classical", "Dance & Electronic", "Blues","Jazz",
|
66 |
+
"Country","Bebop","Folk","Reggae","R&B","Punk","Metal","Pop"]
|
67 |
+
|
68 |
+
xmulti_class_names = ["Metal", "Blues", "Reggae", "Jazz", "Rock", "Folk", "Classical", "Dance & Electronic",
|
69 |
+
"Punk","Bebop", "Pop", "R&B", "Country", "Rap & Hip-Hop", "Soul"]
|
70 |
+
class_indices = {i: class_name for i, class_name in enumerate(class_names)}
|
71 |
+
|
72 |
+
features_list = audio_splitting.split_audio(uploaded_file)
|
73 |
+
features = feature_extraction.scale(features_list)
|
74 |
+
|
75 |
+
st.write(features)
|
76 |
+
# Reshape the features to match the expected shape for prediction
|
77 |
+
reshaped_features = features.reshape(1, -1)
|
78 |
+
if model_name == "XGB - (Multi Label)":
|
79 |
+
# Predict labels for the input features
|
80 |
+
predicted_indices = model.predict(reshaped_features)
|
81 |
+
print(predicted_indices)
|
82 |
+
predicted_labels = []
|
83 |
+
for i in range(0,len(predicted_indices[0])):
|
84 |
+
if predicted_indices[0][i]==1.0:
|
85 |
+
predicted_labels.append(xgb_multi_class_names[i])
|
86 |
+
if predicted_labels:
|
87 |
+
st.write(f"Predicted Genres: {', '.join(predicted_labels)}")
|
88 |
+
else:
|
89 |
+
st.write("No genres predicted for this input.")
|
90 |
+
if model_name == "XGB Classifier - (Single Label)":
|
91 |
+
predicted_indices = model.predict(reshaped_features)
|
92 |
+
predicted_labels = [class_indices[i] for i in predicted_indices]
|
93 |
+
st.write(f"Predicted Genre: {predicted_labels[0]}")
|
94 |
+
elif model_name == "Convolutional Recurrent Neural Network - (Multi Label)"\
|
95 |
+
or model_name == "Neural Network - (Multi Label)"\
|
96 |
+
or model_name == "Batch Normalization - (Multi Label)":
|
97 |
+
predicted_probabilities = model.predict(reshaped_features)
|
98 |
+
|
99 |
+
# Set a threshold for class prediction (e.g., 0.5)
|
100 |
+
threshold = 0.3
|
101 |
+
print(predicted_probabilities)
|
102 |
+
probabilities = []
|
103 |
+
if model_name == "Convolutional Recurrent Neural Network - (Multi Label)":
|
104 |
+
predicted_labels = [class_name for i, class_name in enumerate(multi_class_names) if
|
105 |
+
predicted_probabilities[0][i] >= threshold]
|
106 |
+
probabilities = [(class_name,predicted_probabilities[0][i]*100) for i, class_name in enumerate(multi_class_names)]
|
107 |
+
|
108 |
+
else:
|
109 |
+
predicted_labels = [class_name for i,class_name in enumerate(xmulti_class_names) if
|
110 |
+
predicted_probabilities[0][i] >= threshold]
|
111 |
+
probabilities = [(class_name,predicted_probabilities[0][i]*100) for i, class_name in enumerate(xmulti_class_names)]
|
112 |
+
|
113 |
+
if predicted_labels:
|
114 |
+
st.write(f"All probabilities are:")
|
115 |
+
st.write(probabilities)
|
116 |
+
st.write(f"Predicted Genres: {', '.join(predicted_labels)}")
|
117 |
+
else:
|
118 |
+
st.write("No genre predicted above the threshold.")
|
119 |
+
else:
|
120 |
+
predicted_label = model.predict(reshaped_features)[0]
|
121 |
+
st.write(f"Predicted Genre: {predicted_label}")
|