Update get_track_BOMBYX_7.py

get_track_BOMBYX_7.py

+
+import matplotlib.pyplot as plt
+import pandas as pd
+import scipy.signal as signal
+import soundfile as sf
+from scipy.io import wavfile
+from matplotlib.widgets import Button
+import numpy as np
+import os
+import time
+from wavfile import *
+from tqdm import tqdm
+from audio_v2 import * #Jan code
+from pydub import AudioSegment
+from pydub.playback import play
+import sounddevice as sd
+#load df
+print('loading df... ', end='')
+df = pd.read_pickle('/home/glotinh/Documents/bombyx/dist2/results/clickfiles_maxence/dataframes/BOMBYX7_part/bombyx7_part10')
+convert = lambda fn : pd.to_datetime(fn, format="%Y%m%d_%H%M%S_UTC_BOMBYX 7.tdoa", exact=False).strftime("%Y-%m-%d_%H0000_UTC/%Y%m%d_%H%M%S_UTC_BOMBYX 7.wav")
+
+df['folder_ok']="/home/glotinh/Documents/bombyx/dist2/BOMBYX7"
+
+print('done')
+
+#cast columns into floats
+df["folder_2"]=pd.to_datetime(df.file, format="%Y%m%d_%H%M%S_UTC_BOMBYX 7.tdoa", exact=False).dt.strftime("%Y-%m-%d_%H0000_UTC/%Y%m%d_%H%M%S_UTC_BOMBYX 7.wav")
+df["folder_final"]=df.folder_ok+"/"+df.folder_2
+song= WavFile(df.folder_final.iloc[0])
+rate=song.sample_rate
+fs=song.samples
+sos = signal.butter(5, 5000*2/rate, 'hp', output='sos')
+
+df.time = df.time.astype(float)/rate
+df.argmax = df.argmax.astype(float)
+df.centroid = df.centroid.astype(float)
+df.ipi = df.ipi.astype(float)
+df.tdoa = df.tdoa.astype(float)
+#df.duration = df.duration.astype(float)
+df.sal_pip = df.sal_pip.astype(float)
+df.energy = df.energy.astype(float)
+df["duration"]=900
+#sort ordered by filename == by date
+df.sort_values('file', inplace=True)
+
+#fileDurations = [df[df.filename==fn].iloc[0].duration + 5 for fn in filenames]
+grped = df.groupby('file')
+
+#build array of cumulated durations
+# same array but with the length of the df
+allfilestart = []
+filestarts = [0]
+for i, (fn, gp) in enumerate(grped):
+    filestarts.append(filestarts[-1] + gp.duration.iloc[0]+5)
+    allfilestart.append(np.array(gp.time) + filestarts[i])
+    # allfilestart.extend( [filestarts[i]] * len(df[df.filename==fn]) )
+
+filestarts = filestarts[:-1]
+
+df['abs_click_time'] = np.concatenate(allfilestart, 0)
+
+shortfiles = [f.split('/')[-1][:15] for f in df.file.unique()]
+#extract top decile
+#topdecile = df[ pd.qcut(df.amplitude, 10, labels=False) == 9 ]
+
+# get session sampling rate and create HP filter
+#df["path"]=df.folder +'/'+ df.file
+
+
+
+fig, ax = plt.subplots(2)
+ax[0].scatter(df.abs_click_time, df.tdoa, 1, picker=5)
+ax[0].set_xticks(filestarts)
+ax[0].set_xticklabels(shortfiles)
+plt.xticks(rotation=30)
+#ligne bleu
+for f in filestarts:
+    ax[0].axvline(f, alpha=.5)
+
+winsize = 2.5 # in seconds
+class ppp(object):
+    def __init__(self):
+        self.found=None
+    def onclick(self, event):
+        #get row
+        self.found = df.iloc[np.random.choice(event.ind)]
+
+        ax[0].set_title(self.found.file+'\n Amplitude: '+str(self.found.energy)+' IPI: '+str(self.found.ipi)+' IPI Salience: '+str(self.found.sal_pip)+
+            ' \n Spec amax: '+str(self.found.argmax)+' Spec centr:'+str(self.found.centroid))
+        ax[0].scatter(self.found.abs_click_time, self.found.tdoa, c='r')
+
+        #get click signal, and filter
+        filename = self.found.file[1:] if self.found.file.startswith('../') else self.found.file
+
+        #fs, sig = wavfile.read('../dist2/'+filename, mmap=True)
+        #song= WavFile(self.found.folder_final)
+        print("read wafile")
+        Song=WavFile("/home/glotinh/Documents/bombyx/dist2/BOMBYX7/"+convert(self.found.file))
+        print("on recup le sample rate")
+        rate=Song.sample_rate
+        #print("on recup les samples")
+        #fs=Song.samples
+        print("on recup la pos du clic")
+        clickidx = int(self.found.time * rate)
+        print("on selectionne le clic")
+        sig = Song[clickidx-int(winsize*.5*rate):clickidx+int(winsize*.5*rate),0].copy() # select the click !
+        print("le clic est selectionné")
+        #filtered = signal.sosfilt(sos, sig)
+        ax[1].clear()
+        #ax[1].plot(filtered)
+        ax[1].specgram(sig,Fs=rate)
+        #lire le sons
+        print("on lis le sons")
+        sd.play(sig,rate)
+        print("le sons est lu")
+        fig.canvas.draw()
+
+    def yes(self, event):
+        print(self.found.name)
+        df.at[self.found.name,'labels']='cachcach'
+
+    def no(self, event):
+
+        df.at[self.found.name,'labels']='botbot'
+
+    def glob(self, event):
+
+        df.at[self.found.name,'labels']='globi'
+
+df["labels"]=None
+
+callback = ppp()
+
+
+axyes = plt.axes([0.7, 0.05, 0.1, 0.075])
+axno = plt.axes([0.81, 0.05, 0.1, 0.075])
+axglob = plt.axes([0.5, 0.05, 0.1, 0.075])
+
+bnext = Button(axyes, 'is cachcach')
+bnext.on_clicked(callback.yes)
+
+bprev = Button(axno, 'is boat')
+bprev.on_clicked(callback.no)
+
+bglob = Button(axglob, 'is globi')
+bglob.on_clicked(callback.glob)
+
+
+cid = fig.canvas.mpl_connect('pick_event', callback.onclick)
+plt.show()
+df.to_pickle('BOMBYX_2017_BOMBYX7_part10_tes.pkl')
+
+# other mask
+mask = ((df.centroid>11e3)&(df.centroid<17e3)
+       &(df.ipi>2e-3)&(df.ipi<10e-3)
+       &(df.tdoa>-1.2e-3)&(df.tdoa<1.2e-3)
+       &(df.argmax<20e3))