Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,77 +1,78 @@
|
|
1 |
import streamlit as st
|
|
|
2 |
from tensorflow.keras.models import load_model
|
3 |
from tensorflow.keras.preprocessing.sequence import pad_sequences
|
4 |
import joblib
|
5 |
import pandas as pd
|
6 |
-
|
7 |
from sklearn.preprocessing import StandardScaler
|
|
|
8 |
|
9 |
-
# Load the
|
10 |
-
df = pd.read_csv('df1.csv')
|
11 |
-
df = df.drop(['Unnamed: 0', 'lyrics_filename', 'analysis_url', 'track_href', "type", "id", "uri"], axis=1)
|
12 |
-
|
13 |
-
# Load the tokenizer (ensure it's the one used during training)
|
14 |
-
tokenizer = joblib.load('tokenizer.pkl')
|
15 |
-
|
16 |
-
# Load the emotion prediction model
|
17 |
emotion_model = load_model('lstm_model.h5')
|
18 |
|
19 |
-
#
|
20 |
-
|
21 |
-
'acousticness', 'instrumentalness', 'liveness', 'valence', 'tempo',
|
22 |
-
'duration_ms', 'time_signature']
|
23 |
-
|
24 |
-
audio_features = df[audio_feature_columns]
|
25 |
-
mood_cats = df[['mood_cats']]
|
26 |
-
mood_cats_df = pd.DataFrame(mood_cats)
|
27 |
-
|
28 |
-
# Normalize audio features for content-based
|
29 |
-
scaler_cb = StandardScaler()
|
30 |
-
audio_features_scaled_cb = scaler_cb.fit_transform(audio_features)
|
31 |
-
audio_features_df_cb = pd.DataFrame(audio_features_scaled_cb, columns=audio_feature_columns)
|
32 |
-
combined_features = pd.concat([mood_cats, audio_features_df_cb], axis=1)
|
33 |
|
34 |
-
# Load the
|
35 |
-
|
36 |
|
37 |
-
# Load the
|
38 |
-
|
39 |
|
40 |
-
#
|
41 |
scaler_knn = StandardScaler()
|
42 |
-
audio_features_scaled_knn = scaler_knn.fit_transform(audio_features)
|
43 |
-
audio_features_df_knn = pd.DataFrame(audio_features_scaled_knn, columns=audio_feature_columns)
|
44 |
-
combined_features = pd.concat([mood_cats_df, audio_features_df_knn], axis=1)
|
45 |
-
|
46 |
-
# Load the KNN model
|
47 |
-
knn = joblib.load('knn_model.joblib')
|
48 |
|
49 |
-
#
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
57 |
|
58 |
# Set up the title of the app
|
59 |
-
st.title('
|
60 |
-
|
61 |
-
# Get data from index 0
|
62 |
-
query_data = df.iloc[0]
|
63 |
-
|
64 |
-
# Process the lyrics
|
65 |
-
sequence = tokenizer.texts_to_sequences([query_data['lyrics']])
|
66 |
-
padded_sequence = pad_sequences(sequence, maxlen=50)
|
67 |
-
emotion = emotion_model.predict(padded_sequence).flatten()
|
68 |
|
69 |
-
#
|
70 |
-
|
71 |
|
72 |
-
#
|
73 |
-
hybrid_recs = hybrid_recommendation(
|
74 |
|
75 |
-
|
76 |
-
st.
|
77 |
-
|
|
|
|
1 |
import streamlit as st
|
2 |
+
import numpy as np
|
3 |
from tensorflow.keras.models import load_model
|
4 |
from tensorflow.keras.preprocessing.sequence import pad_sequences
|
5 |
import joblib
|
6 |
import pandas as pd
|
7 |
+
from sklearn.neighbors import NearestNeighbors
|
8 |
from sklearn.preprocessing import StandardScaler
|
9 |
+
from sklearn.metrics.pairwise import cosine_similarity
|
10 |
|
11 |
+
# Load the LSTM model for emotion prediction
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
emotion_model = load_model('lstm_model.h5')
|
13 |
|
14 |
+
# Load the KNN model
|
15 |
+
knn_model = joblib.load('knn_model.joblib')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
16 |
|
17 |
+
# Load the tokenizer
|
18 |
+
tokenizer = joblib.load('tokenizer.pkl')
|
19 |
|
20 |
+
# Load the dataset
|
21 |
+
df = pd.read_csv('df1.csv')
|
22 |
|
23 |
+
# Load the scaler for KNN
|
24 |
scaler_knn = StandardScaler()
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
|
26 |
+
# Function for hybrid recommendation
|
27 |
+
def hybrid_recommendation(song_index):
|
28 |
+
# Get data for the query song
|
29 |
+
query_data = df.iloc[song_index]
|
30 |
+
|
31 |
+
# Process the lyrics for emotion prediction using LSTM
|
32 |
+
sequence = tokenizer.texts_to_sequences([query_data['lyrics']])
|
33 |
+
padded_sequence = pad_sequences(sequence, maxlen=50)
|
34 |
+
predicted_emotion = emotion_model.predict(padded_sequence).flatten()
|
35 |
+
|
36 |
+
# Preprocess for KNN
|
37 |
+
audio_features_knn = query_data[['danceability', 'energy', 'key', 'loudness', 'mode', 'speechiness',
|
38 |
+
'acousticness', 'instrumentalness', 'liveness', 'valence', 'tempo',
|
39 |
+
'duration_ms', 'time_signature']].values.reshape(1, -1)
|
40 |
+
mood_cats = query_data[['mood_cats']]
|
41 |
+
mood_cats_df = pd.DataFrame(mood_cats)
|
42 |
+
audio_features_scaled_knn = scaler_knn.fit_transform(audio_features_knn)
|
43 |
+
combined_features = pd.concat([mood_cats_df, pd.DataFrame(audio_features_scaled_knn, columns=audio_features_knn.columns)], axis=1)
|
44 |
+
|
45 |
+
# Predict using the KNN model
|
46 |
+
knn_recommendations = knn_model.kneighbors(combined_features, n_neighbors=5, return_distance=False)[0]
|
47 |
+
|
48 |
+
# Mapping emotion predictions to encoded categories
|
49 |
+
emotion_mapping = {0: 'happy', 1: 'sad', 2: 'calm', 3: 'anger'}
|
50 |
+
encoded_emotion = np.argmax(predicted_emotion)
|
51 |
+
emotion_category = emotion_mapping[encoded_emotion]
|
52 |
+
|
53 |
+
# Compute cosine similarity for content-based recommendation
|
54 |
+
features_for_similarity = df[['danceability', 'energy', 'key', 'loudness', 'mode', 'speechiness',
|
55 |
+
'acousticness', 'instrumentalness', 'liveness', 'valence', 'tempo',
|
56 |
+
'duration_ms', 'time_signature']].values
|
57 |
+
|
58 |
+
cosine_similarities = cosine_similarity([emotion_category], features_for_similarity).flatten()
|
59 |
+
|
60 |
+
# Combine recommendations from both models
|
61 |
+
combined_indices = np.argsort(-np.concatenate([knn_recommendations, cosine_similarities]))
|
62 |
+
hybrid_recs_sorted = combined_indices[:5] # Select top 5 recommendations
|
63 |
+
|
64 |
+
return hybrid_recs_sorted
|
65 |
|
66 |
# Set up the title of the app
|
67 |
+
st.title('Hybrid Recommender App')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
68 |
|
69 |
+
# Get song index from user input
|
70 |
+
song_index_to_recommend = st.number_input('Enter song index:', min_value=0, max_value=len(df)-1, value=0)
|
71 |
|
72 |
+
# Get hybrid recommendations
|
73 |
+
hybrid_recs = hybrid_recommendation(song_index_to_recommend)
|
74 |
|
75 |
+
# Display the recommendations
|
76 |
+
st.write("Hybrid Recommendations:")
|
77 |
+
for index in hybrid_recs:
|
78 |
+
st.write(f"Song Index: {index}, Title: {df.iloc[index]['title']}, Artist: {df.iloc[index]['artist']}")
|