-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathrhythm.py
230 lines (186 loc) · 8.13 KB
/
rhythm.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
import numpy as np
import essentia.standard as e
class Rhythm(object):
def __init__(self, fname):
self.fpath = fname
self.signal = None #__get_signal__
self.signal_length = None #__get_signal__
self.is_rec = False
self.duration = None #__get_duration__
self.bpm = None #__get_bpm__
self.onset_candidates = [] #onset_candidates
self.frames = list() #onset_candidates
self.frame_count = None #onset_candidates
self.onset_times = None #onset_decider
self.onset_frames = list() #onset_decider
self.onset_count = None #onset_decider
self.onset_energies = list() #__get_onset_energies__
## end of fetching data from original signal
##creating useful data for calculations
self.onset_signal = None
self.shifted_onset_times = []
self.shifted_onset_signal = []
self.stretched_onset_times = []
self.stretched_onset_signal = []
self.final_onset_times = []
self.final_onset_signal = []
self.__get_signal__()
self.__get_duration__()
self.__get_bpm__()
self.__onset_candidate_detection__()
self.__onset_decider__(0.20)
def reset(self):
self.shifted_onset_times = []
self.shifted_onset_signal = []
self.stretched_onset_times = []
self.stretched_onset_signal = []
self.final_onset_times = []
self.final_onset_signal = []
def set_rec(self, ref, mode):
self.is_rec = True
try:
self.__rec_onset_decider__(ref)
self.__shift_times__(ref)
if mode == 0:
self.__stretch_cumulative__(ref)
#print("Stretching cumulatively.")
elif mode == 1:
self.__stretch_adaptive__(ref)
#print("Stretching adaptively.")
else:
pass
except Exception:
print("ERROR: Recording signal is not good.", self.fpath)
return -1
def __get_signal__(self):
"""
:rtype: returns the audio signal by reading the file
"""
e_monoloader = e.MonoLoader(filename=self.fpath)
self.signal = e_monoloader()
self.signal_length = len(self.signal)
def __get_duration__(self):
e_duration = e.Duration()
self.duration = e_duration(self.signal)
def __get_bpm__(self):
e_rhythmextractor2013 = e.RhythmExtractor2013(maxTempo=120, minTempo=40)
bpm, ticks, confidence, estimates, bpmintervals = e_rhythmextractor2013(self.signal)
#print("bpm:", bpm)
assert isinstance(bpm, object)
self.bpm = bpm
def __onset_candidate_detection__(self):
spectrum = e.Spectrum()
e_onsetdetection = e.OnsetDetection(method="flux")
onsetspecs = []
for frame in e.FrameGenerator(self.signal, 1024, 512):
self.frames.append(frame)
onsetspecs.append(spectrum(frame))
self.onset_candidates.append(e_onsetdetection(onsetspecs[-1], [0]*len(onsetspecs[-1])))
self.frame_count = len(self.frames)
def __get_onset_frames__(self):
self.onset_frames = [int((len(self.frames) - 1) * t / self.duration) for t in self.onset_times]
#print(self.onset_frames)
#return onsetframes
def __get_onset_energies__(self):
self.onset_energies = []
e_energy = e.Energy()
englist = list()
for frame in self.onset_frames:
#print(frame, len(self.frames))
eng = e_energy(self.frames[frame])
englist.append(eng)
#print("onset energies initial:", englist)
maxeng = max(englist)
for eng in englist:
if eng > maxeng/2:
self.onset_energies.append(1)
else:
self.onset_energies.append(0.5)
#englist = [eng / maxeng for eng in englist]
#print("onset energies normalized:", self.onset_energies)
#self.onset_energies = englist
def __time2signal__(self, timelist):
SR = 44100
#tempbeatsignal = [0] * self.length
tempbeatsignal = [0] * int(1 + max(timelist) * SR)
cnt = 0
for x in timelist:
#print(int(x * SR), len(tempbeatsignal))
tempbeatsignal[int(x * SR)] = self.onset_energies[cnt] #temporarily disabled for generating figures
#tempbeatsignal[int(x * SR)] = 1
cnt += 1
#self.onset_signal = tempbeatsignal
return tempbeatsignal
def __onset_decider__(self, noise_threshold):
e_onsets = e.Onsets(silenceThreshold=noise_threshold, frameRate=44100 / 512.0)
onsetdetectsM = [self.onset_candidates]
onsetresults = e_onsets(onsetdetectsM, [1])
self.onset_times = onsetresults
self.onset_count = len(self.onset_times)
self.__get_onset_frames__()
self.__get_onset_energies__()
self.onset_signal = self.__time2signal__(self.onset_times)
def __rec_onset_decider__(self, ref):
nof_tries = 0
threshold = 0.15
while self.onset_count != ref.onset_count:
if self.onset_count < ref.onset_count:
#print("Decrease threshold.", threshold)
if threshold - 0.02 <= 0:
print("Signal not good.")
break
else:
threshold -= 0.02
#self.performance = audionset.AudioPiece(WAVE_OUTPUT_FILENAME, threshold)
else:
threshold += 0.035
#print("Increase threshold.", threshold)
self.__onset_decider__(threshold)
nof_tries += 1
if nof_tries == 50:
print("Cannot detect good number of onsets from recording.")
break
#print("Threshold:", threshold)
def __shift_times__(self, ref):
#CALL JUST AFTER ONSET DETECTION
amount = float(self.onset_times[0])
#print(amount)
#print("Rhythm onset times length:", len(self.onset_times))
for e in self.onset_times:
self.shifted_onset_times.append(e - amount + ref.onset_times[0])
#print("Rhythm shifted onset times:", self.shifted_onset_times)
self.shifted_onset_signal = self.__time2signal__(self.shifted_onset_times)
self.shifted_onset_signal = self.shifted_onset_signal[:1 + max([i for i, j in enumerate(self.shifted_onset_signal) if j == 1])]
def __stretch_cumulative__(self, ref):
#self.shifttimes(ref)
ratio = (ref.onset_times[-1] - ref.onset_times[0]) / (self.shifted_onset_times[-1] - self.shifted_onset_times[0])
#print("Stretch ratio=", ratio)
for i in self.shifted_onset_times:
self.stretched_onset_times.append((i - self.shifted_onset_times[0]) * ratio + self.shifted_onset_times[0])
self.stretched_onset_signal = self.__time2signal__(self.stretched_onset_times)
self.stretched_onset_signal = self.stretched_onset_signal[:ref.signal_length]
def __stretch_adaptive__(self, ref):
distances = list()
distancesums = list()
mindistance = 100
ratio = 0.5
rate = 0.01
x = ref.onset_times
y = self.shifted_onset_times
for i in range(0, 301):
scaled_y = ratio * (y - y[0]) + y[0]
#print("Scaler:", i, "Scaled Y:", scaled_y)
distances.append(x - scaled_y)
distancesums.append(sum(abs(distances[-1])))
ratio += rate
mindistance = min(distancesums)
index_of_mindistance = distancesums.index(mindistance)
closest_scaled_y = distances[index_of_mindistance]
ratio = 0.5 + (index_of_mindistance * rate)
#print(mindistance, index_of_mindistance, ratio)
#print("Mind difference:", closest_scaled_y)
y = ratio * (y - y[0]) + y[0]
#print("Final Y:", y)
self.stretched_onset_times = y
self.stretched_onset_signal = self.__time2signal__(self.stretched_onset_times)
self.stretched_onset_signal = self.stretched_onset_signal[:ref.signal_length]