Noise Reduction Script (avinashkranjan#1106)

Author: A-kriti

Noise Reduction Script/Audio/(Filtered_Audio)test_noise .wav

@@ -0,0 +1,50 @@
+# Spectral Subtraction: Method used for noise reduction 
+import scipy.io.wavfile as wav
+import numpy as np
+import matplotlib.pyplot as plt
+
+file = input("Enter the file path: ")
+sr, data = wav.read(file)
+fl = 400 #frame_length
+frames = []  #empty list 
+for i in range(0,int(len(data)/(int(fl/2))-1)):
+    arr = data[int(i*int(fl/2)):int(i*int(fl/2)+fl)]  
+    frames.append(arr)    #appending each array data into the frames list
+frames = np.array(frames)   #converting the frames list into an array
+ham_window = np.hamming(fl)    #using np.hamming 
+windowed_frames = frames*ham_window    #multiplying frames array with ham_window
+dft = []   #empty list containing fft of windowed_frames
+for i in windowed_frames:
+    dft.append(np.fft.fft(i))   #now taking the first fourier transform of each window
+dft = np.array(dft)  #converting dft into array
+
+dft_mag_spec = np.abs(dft)      #converting dft into absolute values
+dft_phase_spec = np.angle(dft)   #finding dft angle
+noise_estimate = np.mean(dft_mag_spec,axis=0)    #mean
+noise_estimate_mag = np.abs(noise_estimate)   #absolute value
+
+estimate_mag = (dft_mag_spec-2*noise_estimate_mag)   #subtraction method
+estimate_mag[estimate_mag<0]=0
+estimate = estimate_mag*np.exp(1j*dft_phase_spec)  #calculating the final estimate
+ift = []   #taking ift as input list containing inverse fourier transform of estimate
+for i in estimate:
+    ift.append(np.fft.ifft(i))    #appending in ift list 
+
+clean_data = []
+clean_data.extend(ift[0][:int(fl/2)])     #extending clean_data containg ift list
+for i in range(len(ift)-1):   
+    clean_data.extend(ift[i][int(fl/2):]+ift[i+1][:int(fl/2)])
+clean_data.extend(ift[-1][int(fl/2):])   #extending clean_data containing ift list
+clean_data = np.array(clean_data)  #converting it into array
+
+#finally plotting the graph showing the diffrence in the noise
+fig = plt.figure(figsize=(8,5))
+ax = plt.subplot(1,1,1)
+ax.plot(np.linspace(0,64000,64000),data,label='Original',color="orange")
+ax.plot(np.linspace(0,64000,64000),clean_data,label='Filtered',color="purple")
+ax.legend(fontsize=12)
+ax.set_title('Spectral Subtraction Method', fontsize=15)
+filename = os.path.basename(file)
+cleaned_file = "(Filtered_Audio)"+filename   #final filtered audio
+wav.write(cleaned_file,rate=sr, data = clean_data.astype(np.int16))
+plt.savefig(filename+"(Spectral Subtraction graph).jpg")  #saved file name as audio.wav(Spectral Subtraction graph).jpg

Noise Reduction Script/Audio/test_noise.wav

@@ -0,0 +1,25 @@
+# Noise Reduction Script
+Implementing a feature that helps to filter an audio file by reducing the background noise similar to "Audacity".
+
+## Libraries used
+Firstly import the following python libraries 
+* NumPy
+* scipy.io.wavfile
+* Matplotlib
+* Os
+Save the audio files and your code in the same folder
+Run the python code
+
+## Detailed explanation of method used for "Noise Reduction Script" 
+* Imported the required libraries (NumPy, scipy.io.wavfile, and Matplotlib)
+* Read the input audio file using scipy.io.wavfile library
+* Converting the audio file into an array containg all the information of the given audio file and intiallizing the frame value.
+* Calculating the first fourier transform of each window of the noisy audio file 
+* Subtracting the noise spectral mean from input spectral, and istft (Inverse Short-Time Fourier Transform)
+* Finally getting an audio file with reduction in the background noise at a much higher extent
+
+## Output
+<img src="Graph/test_noise.wav(Spectral Subtraction graph).jpg" height="500px">
+
+## Author(s)
+[Akriti](https://github.com/A-kriti)

Noise Reduction Script/Graph/test_noise.wav(Spectral Subtraction graph).jpg

@@ -0,0 +1,3 @@
+matplotlib==3.2.2
+numpy==1.18.5
+scipy==1.5.0