added display to use the mel-scale approximation on the stochastic an…

…alysis, thus supporting both linear frequency and mel-scale approximations

software/models_interface/stochasticModel_GUI_frame.py

@@ -65,15 +65,33 @@ def initUI(self):
         self.stocf.delete(0, END)
         self.stocf.insert(0, "0.1")
 
+        #MEl SCALE
+        melScale_label = "Approximation scale (0: linear, 1: mel):"
+        Label(self.parent, text=melScale_label).grid(row=5, column=0, sticky=W, padx=5, pady=(10,2))
+        self.melScale = Entry(self.parent, justify=CENTER)
+        self.melScale["width"] = 5
+        self.melScale.grid(row=5, column=0, sticky=W, padx=(285,5), pady=(10,2))
+        self.melScale.delete(0, END)
+        self.melScale.insert(0, "1")
+
+        #NORMALIZATION
+        normalization_label = "Amplitude normalization (0: no, 1: yes):"
+        Label(self.parent, text=normalization_label).grid(row=6, column=0, sticky=W, padx=5, pady=(10,2))
+        self.normalization = Entry(self.parent, justify=CENTER)
+        self.normalization["width"] = 5
+        self.normalization.grid(row=6, column=0, sticky=W, padx=(285,5), pady=(10,2))
+        self.normalization.delete(0, END)
+        self.normalization.insert(0, "1")
+
         #BUTTON TO COMPUTE EVERYTHING
         self.compute = Button(self.parent, text="Compute", command=self.compute_model)
-        self.compute.grid(row=5, column=0, padx=5, pady=(10,2), sticky=W)
+        self.compute.grid(row=7, column=0, padx=5, pady=(10,2), sticky=W)
 
         #BUTTON TO PLAY OUTPUT
         output_label = "Stochastic:"
-        Label(self.parent, text=output_label).grid(row=6, column=0, sticky=W, padx=5, pady=(10,15))
+        Label(self.parent, text=output_label).grid(row=8, column=0, sticky=W, padx=5, pady=(10,15))
         self.output = Button(self.parent, text=">", command=lambda:UF.wavplay('output_sounds/' + os.path.basename(self.filelocation.get())[:-4] + '_stochasticModel.wav'))
-        self.output.grid(row=6, column=0, padx=(80,5), pady=(10,15), sticky=W)
+        self.output.grid(row=8, column=0, padx=(80,5), pady=(10,15), sticky=W)
 
         # define options for opening file
         self.file_opt = options = {}
@@ -97,8 +115,10 @@ def compute_model(self):
             H = int(self.H.get())
             N = int(self.N.get())
             stocf = float(self.stocf.get())
+            melScale = int(self.melScale.get())
+            normalization = int(self.normalization.get())
 
-            stochasticModel_function.main(inputFile, H, N, stocf)
+            stochasticModel_function.main(inputFile, H, N, stocf, melScale, normalization)
 
         except ValueError as errorMessage:
             messagebox.showerror("Input values error", errorMessage)

software/models_interface/stochasticModel_function.py

@@ -3,26 +3,32 @@
 import numpy as np
 import matplotlib.pyplot as plt
 import os, sys
-from scipy.signal import get_window
+from scipy.signal.windows import hann
 sys.path.append(os.path.join(os.path.dirname(os.path.realpath(__file__)), '../models/'))
 import utilFunctions as UF
 import stochasticModel as STM
+import stft as STFT
 
-def main(inputFile='../../sounds/ocean.wav', H=256, N=512, stocf=.1):
+def main(inputFile='../../sounds/ocean.wav', H=256, N=512, stocf=.1, melScale=1, normalization=1):
 	"""
 	inputFile: input sound file (monophonic with sampling rate of 44100)
 	H: hop size, N: fft size
 	stocf: decimation factor used for the stochastic approximation (bigger than 0, maximum 1)
+	melScale: frequency approximation scale (0: linear approximation, 1: mel frequency approximation)
+	normalization: amplitude normalization of output (0: no normalization, 1: normalization to input amplitude)
 	"""
 
 	# read input sound
 	(fs, x) = UF.wavread(inputFile)
 
 	# compute stochastic model
-	stocEnv = STM.stochasticModelAnal(x, H, N, stocf)
+	stocEnv = STM.stochasticModelAnal(x, H, N, stocf, fs, melScale)
 
 	# synthesize sound from stochastic model
-	y = STM.stochasticModelSynth(stocEnv, H, N)
+	y = STM.stochasticModelSynth(stocEnv, H, N, fs, melScale)
+
+	if (normalization==1):
+		y = y * max(x)/max(y)
 
 	outputFile = 'output_sounds/' + os.path.basename(inputFile)[:-4] + '_stochasticModel.wav'
 
@@ -32,31 +38,32 @@ def main(inputFile='../../sounds/ocean.wav', H=256, N=512, stocf=.1):
 	# create figure to plot
 	plt.figure(figsize=(9, 6))
 
-	# plot the input sound
-	plt.subplot(3,1,1)
-	plt.plot(np.arange(x.size)/float(fs), x)
-	plt.axis([0, x.size/float(fs), min(x), max(x)])
-	plt.ylabel('amplitude')
-	plt.xlabel('time (sec)')
-	plt.title('input sound: x')
+	# frequency range to plot
+	maxplotfreq = 10000.0
 
-	# plot stochastic representation
-	plt.subplot(3,1,2)
-	numFrames = int(stocEnv[:,0].size)
-	frmTime = H*np.arange(numFrames)/float(fs)  
-	binFreq = np.arange(int(stocf*(N/2+1)))*float(fs)/(stocf*N)                      
-	plt.pcolormesh(frmTime, binFreq, np.transpose(stocEnv), shading='auto')
-	plt.autoscale(tight=True)
+	# plot input spectrogram
+	plt.subplot(2,1,1)
+	mX, pX = STFT.stftAnal(x, hann(N), N, H)
+	numFrames = int(mX[:,0].size)
+	frmTime = H*np.arange(numFrames)/float(fs)
+	binFreq = fs*np.arange(N*maxplotfreq/fs)/N
+	plt.pcolormesh(frmTime, binFreq, np.transpose(mX[:,:int(N*maxplotfreq/fs+1)]))
 	plt.xlabel('time (sec)')
 	plt.ylabel('frequency (Hz)')
-	plt.title('stochastic approximation')
+	plt.title('input magnitude spectrogram')
+	plt.autoscale(tight=True)
 
 	# plot the output sound
-	plt.subplot(3,1,3)
-	plt.plot(np.arange(y.size)/float(fs), y)
-	plt.axis([0, y.size/float(fs), min(y), max(y)])
-	plt.ylabel('amplitude')
+	plt.subplot(2,1,2)
+	mY, pY = STFT.stftAnal(y, hann(N), N, H)
+	numFrames = int(mY[:,0].size)
+	frmTime = H*np.arange(numFrames)/float(fs)
+	binFreq = fs*np.arange(N*maxplotfreq/fs)/N
+	plt.pcolormesh(frmTime, binFreq, np.transpose(mY[:,:int(N*maxplotfreq/fs+1)]))
 	plt.xlabel('time (sec)')
+	plt.ylabel('frequency (Hz)')
+	plt.title('input magnitude spectrogram')
+	plt.autoscale(tight=True)
 
 	plt.tight_layout()
 	plt.ion()