[Python] How to Resolve IndexError: tuple index out of range for my python code

In the scope of my projects, i've been trying to make noise filter integrated with Artifical Intelligence. But when compiled the following code, i encountered the results that i specify it below my code. How can i resolve it? Anyone help me?

import librosa
import librosa.display
import numpy as np
import matplotlib.pyplot as plt
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, BatchNormalization, Reshape, Input
from sklearn.model_selection import train_test_split
from scipy.io.wavfile import write

# MFCC özelliklerini çıkarma fonksiyonu
def extract_mfcc(audio, sample_rate, n_mfcc=20, n_fft=2048, hop_length=512):
mfcc = librosa.feature.mfcc(y=audio, sr=sample_rate, n_mfcc=n_mfcc, n_fft=n_fft, hop_length=hop_length)
return mfcc.T # Zaman eksenine uygun hale getir

# CNN modeli oluşturma
def create_cnn_model(input_shape):
model = Sequential([
Input(shape=input_shape),
Conv2D(32, (3, 3), activation='relu', padding="same"),
BatchNormalization(),
MaxPooling2D((2, 2), padding="same"),
Conv2D(64, (3, 3), activation='relu', padding="same"),
BatchNormalization(),
MaxPooling2D((2, 2), padding="same"),
Flatten(),
Dense(128, activation='relu'),
BatchNormalization(),
Dense(np.prod(input_shape[:-1]), activation='linear'),
Reshape(input_shape[:-1]) # Çıkışı orijinal şekle döndür
])
model.compile(optimizer='adam', loss='mse', metrics=['mse'])
return model

# Gürültüsüz ve gürültülü ses dosyalarını yükle
clean_file_path = "denoised.wav"
noisy_file_path = "CPLT_140knt.wav"

clean_audio, sample_rate_clean = librosa.load(clean_file_path, sr=None)
noisy_audio, sample_rate_noisy = librosa.load(noisy_file_path, sr=None)

# Uzunlukları eşitle
min_length = min(len(clean_audio), len(noisy_audio))
clean_audio = clean_audio[:min_length]
noisy_audio = noisy_audio[:min_length]

# MFCC çıkarma
clean_mfcc = extract_mfcc(clean_audio, sample_rate_clean)
noisy_mfcc = extract_mfcc(noisy_audio, sample_rate_noisy)

# Veri seti oluşturma
X = np.expand_dims(noisy_mfcc, axis=-1) # Gürültülü MFCC
y = np.expand_dims(clean_mfcc, axis=-1) # Gürültüsüz MFCC

# Eğitim ve test veri setlerini ayırma
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

# CNN giriş boyutunu belirleme
_, height, width = X_train.shape
channels = 1
input_shape = (height, width, channels)

# CNN modeli oluşturma
model = create_cnn_model(input_shape)

# Modeli eğitme
print("Model eğitiliyor...")
model.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=250, batch_size=128, verbose=1)

# Gürültüyü temizleme
predicted_mfcc = model.predict(X_test)

# MFCC'den ses sinyalini yeniden oluşturma
def reconstruct_audio_from_mfcc(mfcc, sr, n_fft=2048, hop_length=512):
mel_spec = librosa.feature.inverse.mfcc_to_mel(mfcc.T)
mel_spec = np.maximum(mel_spec, 1e-10) # Negatif veya sıfır değerleri önlemek için küçük bir minimum belirle
# Mel-spektrogramdan ses sinyali oluşturma
reconstructed_audio = librosa.feature.inverse.mel_to_audio(mel_spec, sr=sr, n_fft=n_fft, hop_length=hop_length)
return reconstructed_audio

# İlk test örneğini yeniden oluştur
reconstructed_audio = reconstruct_audio_from_mfcc(predicted_mfcc[0, ..., 0], sample_rate_clean)

# Sonuçları görselleştirme
plt.figure(figsize=(15, 5))
plt.subplot(3, 1, 1)
librosa.display.waveshow(noisy_audio, sr=sample_rate_noisy)
plt.title("Gürültülü Ses")
plt.subplot(3, 1, 2)
librosa.display.waveshow(clean_audio, sr=sample_rate_clean)
plt.title("Gürültüsüz Referans Ses")
plt.subplot(3, 1, 3)
librosa.display.waveshow(reconstructed_audio, sr=sample_rate_clean)
plt.title("CNN ile Temizlenmiş Ses")
plt.tight_layout()
plt.show()

# Temizlenmiş sesi kaydetme
output_file = "cnn_denoised_audio.wav"
write(output_file, sample_rate_clean, (reconstructed_audio * 32767).astype(np.int16))
print(f"CNN ile temizlenmiş ses '{output_file}' olarak kaydedildi.")


Terminal Output:

Traceback (most recent call last):
File "c:\Python\Noise Filter.py", line 81, in <module>
reconstructed_audio = reconstruct_audio_from_mfcc(predicted_mfcc[0, ..., 0], sample_rate_clean)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "c:\Python\Noise Filter.py", line 74, in reconstruct_audio_from_mfcc
mel_spec = librosa.feature.inverse.mfcc_to_mel(mfcc.T)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\atabe\AppData\Local\Programs\Python\Python312\Lib\site-packages\librosa\feature\inverse.py", line 278, in mfcc_to_mel
logmel = scipy.fftpack.idct(mfcc, axis=-2, type=dct_type, norm=norm, n=n_mels)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\atabe\AppData\Local\Programs\Python\Python312\Lib\site-packages\scipy\fftpack\_realtransforms.py", line 440, in idct
return _pocketfft.dct(x, type, n, axis, norm, overwrite_x)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\atabe\AppData\Local\Programs\Python\Python312\Lib\site-packages\scipy\fft\_pocketfft\realtransforms.py", line 31, in _r2r
tmp, copied = _fix_shape_1d(tmp, n, axis)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\atabe\AppData\Local\Programs\Python\Python312\Lib\site-packages\scipy\fft\_pocketfft\helper.py", line 149, in _fix_shape_1d
return _fix_shape(x, (n,), (axis,))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\atabe\AppData\Local\Programs\Python\Python312\Lib\site-packages\scipy\fft\_pocketfft\helper.py", line 124, in _fix_shape
if x.shape[ax] >= n:
~~~~~~~^^^^
IndexError: tuple index out of range


I expects to get resolved issues that i specified

Continue reading...