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
231
232
233
234
235
236
237
238
|
from ReadAudio import AudioReader, ConvertReader
import soundtouch, wave, audioop, math
from array import array
class Shifter:
@staticmethod
def shift_chunk(chunk, sampling_rate, channels, shift):
"""Shift the pitch of a chunk of audio up or down
Width must be 2."""
st = soundtouch.SoundTouch(sampling_rate, channels)
st.set_pitch_shift(shift)
ii = 0
resstr = ""
while ii + 4608 < len(chunk):
st.put_samples(chunk[ii:ii+4608].tostring())
ii += 4608
while st.ready_count() > 0:
resstr += st.get_samples(4608)
st.put_samples(chunk[ii:])
while st.ready_count() > 0:
resstr += st.get_samples(11025)
resstr += Shifter.get_flush(st, channels, len(chunk) - len(resstr) / 2)
del st
result = array(chunk.typecode)
result.fromstring(resstr)
return result
@staticmethod
def many_shift_chunk(chunk, sampling_rate, channels, shifts):
"""Produce harmonies by shifting a chunk of audio more than once and combining them."""
shifteds = []
maxlen = 0
for jj in xrange(len(shifts)):
if not shifts[jj]:
shifted = chunk
else:
shifted = Shifter.shift_chunk(chunk, sampling_rate, channels, shifts[jj])
shifteds.append(shifted)
maxlen = max(maxlen, len(shifted))
if len(shifteds) > 1:
newchunk = [0] * maxlen
for ii in xrange(maxlen):
count = 0
for jj in xrange(len(shifteds)):
if len(shifteds[jj]) > ii:
newchunk[ii] += shifteds[jj][ii]
count += 1
newchunk[ii] /= count
result = array(chunk.typecode)
result.fromlist(newchunk)
return result
else:
return shifteds[0]
@staticmethod
def raw_shift_reader(srcpath, dstpath, shift):
"""Shift an entire file up or down"""
# Open the file and convert it to have SoundTouch's required 2-byte samples
reader = AudioReader.open(srcpath)
reader2 = ConvertReader(reader, set_raw_width=2)
# Create the SoundTouch object and set the given shift
st = soundtouch.SoundTouch(reader2.sampling_rate(), reader2.channels())
st.set_pitch_shift(shift)
# Create the .WAV file to write the result to
writer = wave.open(dstpath, 'w')
writer.setnchannels(reader2.channels())
writer.setframerate(reader2.sampling_rate())
writer.setsampwidth(reader2.raw_width())
# Read values and feed them into SoundTouch
while True:
data = reader2.raw_read()
if not data:
break
print len(data)
st.put_samples(data)
while st.ready_count() > 0:
writer.writeframes(st.get_samples(11025))
# Flush any remaining values
writer.writeframes(Shifter.get_flush(st, reader2.channels()))
# Clean up
writer.close()
reader2.close()
@staticmethod
def get_flush(st, channels, fade=0):
"""Like soundtouch's flush, don't require that all data comes through, just any.
If fade > 0, only allow [fade] samples, and linearly scale volume to 0 over that length"""
waiting = st.waiting_count()
ready = st.ready_count()
result = ""
silence = array('h', [0] * 64)
while st.ready_count() == ready:
st.put_samples(silence)
while st.ready_count() > 0:
result += st.get_samples(11025)
st.clear()
if len(result) > 2 * channels * waiting:
result = result[0:(2 * channels * waiting)]
fade = min(fade, len(result) / 2)
if fade > 0:
resultstring = ""
for ii in xrange(fade / channels):
i0 = ii * 2*channels
i1 = (ii+1) * 2*channels
resultstring += audioop.mul(result[i0:i1], 2, 1 - float(ii) / (fade / channels))
result = resultstring
return result
@staticmethod
def bpm_detect_file(fullpath):
"""Detect the beat from an entire file"""
reader = AudioReader.open(fullpath)
reader2 = ConvertReader(reader, set_raw_width=2)
bd = soundtouch.BPMDetect(reader2.sampling_rate(), reader2.channels())
while True:
data = reader2.raw_read()
if not data:
break
bd.put_samples(data)
reader2.close()
return bd.get_bpm()
@staticmethod
def echocancel(outputdata, inputdata):
"""Try to identify an echo and remove it.
Should contain 2-byte samples"""
pos = audioop.findmax(outputdata, 800)
out_test = outputdata[pos*2:]
in_test = inputdata[pos*2:]
ipos, factor = audioop.findfit(in_test, out_test)
factor = audioop.findfactor(in_test[ipos*2:ipos*2+len(out_test)], out_test)
prefill = '\0'*(pos+ipos)*2
postfill = '\0'*(len(inputdata) - len(prefill) - len(outputdata))
outputdata = prefill + audioop.mul(outputdata, 2, -factor) + postfill
return audioop.add(inputdata, outputdata, 2)
@staticmethod
def beats_to_ms(bpm, beats):
"""Convert from bpm at a given beat rate to ms between beats."""
return 60 * 1000 * beats / bpm
@staticmethod
def find_division_start(fullpath, bpm, beats_per):
"""Identify the start of the beats, by finding segments that fit together"""
reader = AudioReader.open(fullpath)
# This doesn't find the exact time of the max, but don't need it.
max_value = 0
max_time = 0
while True:
data = reader.raw_read()
if data is None:
break
data_max = audioop.max(data, reader.raw_width())
if data_max > max_value:
max_value = data_max
max_time = reader.current_time()
before = max_time - Shifter.beats_to_ms(bpm, beats_per)
after = max_time + 2 * Shifter.beats_to_ms(bpm, beats_per)
if before < 0:
after += -before
before = 0
if after > reader.duration():
before -= after - reader.duration()
after = reader.duration()
if before < 0:
if beats_per < 2:
raise RuntimeError('This audio file is too short to be divided by beats.')
else:
reader.close()
return Shifting.find_division_start(filepath, bpm, int(beats_per / 2))
reader.seek_time(0)
reader2 = ConvertReader(reader, set_raw_width=2, set_channels=1)
region = reader2.raw_random_read(before, after)
# both in bytes
raw_length = 2 * int(len(region) / 6)
beat_length = int(2 * Shifter.beats_to_ms(bpm, 1) * reader2.sampling_rate() / 1000.0)
print "Around max: " + str(before) + " - " + str(after) + ": " + str(raw_length)
min_factor = 0
min_ii = 0
# First determine time within a beat
for ii in xrange(beat_length / 200):
factor = audioop.findfactor(region[200*ii:200*ii+raw_length], region[200*ii+raw_length:200*ii+2*raw_length])
if factor < min_factor:
print "Samp: At " + str(ii) + " " + str(factor)
min_factor = factor
min_ii = ii
# Second, determine which beat to use
min_factor = 0
min_jj = 0
for jj in xrange(beats_per):
factor = audioop.findfactor(region[jj*beat_length+200*min_ii:jj*beat_length+200*min_ii+raw_length], region[jj*beat_length+200*min_ii+raw_length:jj*beat_length+200*min_ii+2*raw_length])
print "Beat: At " + str(jj) + " " + str(factor)
if factor < min_factor:
min_factor = factor
min_jj = jj
print "Best: Beat: " + str(min_jj) + ", Samp: " + str(100*min_ii)
start_time = before + (min_jj*beat_length*2 + 100*min_ii) * 1000.0 / reader2.sampling_rate()
reader2.close()
return math.fmod(start_time, Shifter.beats_to_ms(bpm, beats_per))
|
