forked from f90/Wave-U-Net
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Datasets.py
333 lines (269 loc) · 13.7 KB
/
Datasets.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
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
import os.path
import Input.Input
from Sample import Sample
import csv
import numpy as np
from lxml import etree
import librosa
import soundfile
import os
import fnmatch
from exceptions import Exception
import musdb
def subtract_audio(mix_list, instrument_list):
'''
Generates new audio by subtracting the audio signal of an instrument recording from a mixture
:param mix_list:
:param instrument_list:
:return:
'''
assert(len(mix_list) == len(instrument_list))
new_audio_list = list()
for i in range(0, len(mix_list)):
new_audio_path = os.path.dirname(mix_list[i]) + os.path.sep + "remainingmix" + os.path.splitext(mix_list[i])[1]
new_audio_list.append(new_audio_path)
if os.path.exists(new_audio_path):
continue
mix_audio, mix_sr = librosa.load(mix_list[i], mono=False, sr=None)
inst_audio, inst_sr = librosa.load(instrument_list[i], mono=False, sr=None)
assert (mix_sr == inst_sr)
new_audio = mix_audio - inst_audio
if not (np.min(new_audio) >= -1.0 and np.max(new_audio) <= 1.0):
print("Warning: Audio for mix " + str(new_audio_path) + " exceeds [-1,1] float range!")
librosa.output.write_wav(new_audio_path, new_audio, mix_sr) #TODO switch to compressed writing
print("Wrote accompaniment for song " + mix_list[i])
return new_audio_list
def create_sample(db_path, instrument_node):
path = db_path + os.path.sep + instrument_node.xpath("./relativeFilepath")[0].text
sample_rate = int(instrument_node.xpath("./sampleRate")[0].text)
channels = int(instrument_node.xpath("./numChannels")[0].text)
duration = float(instrument_node.xpath("./length")[0].text)
return Sample(path, sample_rate, channels, duration)
def getAllFilesOfType(root_path, extension):
matches = []
for root, dirnames, filenames in os.walk(root_path):
for filename in fnmatch.filter(filenames, '*.' + extension):
matches.append(os.path.join(root, filename))
return matches
def get_samples_in_folder(audio_path, extension):
audio_file_list = getAllFilesOfType(audio_path, extension)
sample_list = list()
for audio_file in audio_file_list:
print("Reading in metadata of file " + audio_file)
try:
sample = Sample.from_path(audio_file)
except Exception:
print("Skipping sample at path " + audio_path)
continue
sample_list.append(sample)
assert(len(sample_list) > 0) # If we did not find any samples something must have gone wrong
return sample_list
def getFullFMA(audio_path):
return get_samples_in_folder(audio_path, "mp3")
def getLooperman(audio_path):
return get_samples_in_folder(audio_path, "mp3")
def getPopFMA(database_path, audio_path=None):
return getFMAGenre(10, database_path, audio_path)
def getVocalFMA(database_path, audio_path=None):
if audio_path is None:
audio_path = database_path
track_csv = os.path.join(database_path, "fma_metadata", "raw_tracks.csv")
sample_list = list()
with open(track_csv, 'rb') as csvfile:
reader = csv.DictReader(csvfile)
for row in reader:
if not int(row["track_instrumental"]):
track_id = int(row["track_id"])
filename = '{:06d}'.format(track_id) + ".mp3"
folder_id = track_id // 1000
foldername = '{:03d}'.format(folder_id)
audio_file = os.path.join(audio_path, "fma_full", foldername, filename)
print("Reading in metadata of file " + audio_file)
try:
sample = Sample.from_path(audio_file)
except Exception as e:
print("Skipping sample at path " + audio_file)
print(e)
continue
sample_list.append(sample)
return sample_list
def convert_float_to_pcm(float_audio):
assert(isinstance(float_audio, np.ndarray))
assert(float_audio.dtype == np.float32 or float_audio.dtype == np.float64)
return (float_audio * 32767).astype(np.int16)
def write_wav_skip_existing(path, y, sr, norm=False):
if not os.path.exists(path):
soundfile.write(path, y, sr, "PCM_16")
else:
print("WARNING: Tried writing audio to " + path + ", but audio file exists already. Skipping file!")
def getMUSDB(database_path):
mus = musdb.DB(root_dir=database_path, is_wav=False)
subsets = list()
for subset in ["train", "test"]:
tracks = mus.load_mus_tracks(subset)
samples = list()
for track in tracks:
# Get mix and instruments
# Bass
bass_path = track.path[:-4] + "_bass.wav"
bass_audio = track.sources["bass"].audio
write_wav_skip_existing(bass_path, bass_audio, track.rate)
bass = Sample(bass_path, track.rate, track.audio.shape[1], track.duration)
# Drums
drums_path = track.path[:-4] + "_drums.wav"
drums_audio = track.sources["drums"].audio
write_wav_skip_existing(drums_path, drums_audio, track.rate)
drums = Sample(drums_path, track.rate, track.audio.shape[1], track.duration)
# Other
other_path = track.path[:-4] + "_other.wav"
other_audio = track.sources["other"].audio
write_wav_skip_existing(other_path, other_audio, track.rate)
other = Sample(other_path, track.rate, track.audio.shape[1], track.duration)
# Vocals
vocal_path = track.path[:-4] + "_vocals.wav"
vocal_audio = track.sources["vocals"].audio
write_wav_skip_existing(vocal_path, vocal_audio, track.rate)
vocal = Sample(vocal_path, track.rate, track.audio.shape[1], track.duration)
# Add other instruments to form accompaniment
acc_audio = track.sources["drums"].audio + track.sources["bass"].audio + track.sources["other"].audio
acc_path = track.path[:-4] + "_accompaniment.wav"
write_wav_skip_existing(acc_path, acc_audio, track.rate)
acc = Sample(acc_path, track.rate, track.audio.shape[1], track.duration)
# Create mixture
mix_path = track.path[:-4] + "_mix.wav"
mix_audio = track.audio
write_wav_skip_existing(mix_path, mix_audio, track.rate)
mix = Sample(mix_path, track.rate, track.audio.shape[1], track.duration)
diff_signal = np.abs(mix_audio - bass_audio - drums_audio - other_audio - vocal_audio)
print("Maximum absolute deviation from source additivity constraint: " + str(np.max(diff_signal)))# Check if acc+vocals=mix
print("Mean absolute deviation from source additivity constraint: " + str(np.mean(diff_signal)))
samples.append((mix, acc, bass, drums, other, vocal)) # Collect all sources for now. Later on for SVS: [mix, acc, vocal] Multi-instrument: [mix, bass, drums, other, vocals]
subsets.append(samples)
return subsets
def getFMAGenre(genre_id, database_path, audio_path=None):
if audio_path is None:
audio_path = database_path
track_csv = os.path.join(database_path, "fma_metadata", "raw_tracks.csv")
sample_list = list()
with open(track_csv, 'rb') as csvfile:
reader = csv.DictReader(csvfile)
for row in reader:
if row["track_genres"].__contains__("genre_id': '" + str(genre_id) + "'"):
track_id = int(row["track_id"])
filename = '{:06d}'.format(track_id) + ".mp3"
folder_id = track_id // 1000
foldername = '{:03d}'.format(folder_id)
audio_file = os.path.join(audio_path, "fma_full", foldername, filename)
print("Reading in metadata of file " + audio_file)
try:
sample = Sample.from_path(audio_file)
except Exception as e:
print("Skipping sample at path " + audio_file)
print(e)
continue
sample_list.append(sample)
return sample_list
def getDSDFilelist(xml_path):
tree = etree.parse(xml_path)
root = tree.getroot()
db_path = root.find("./databaseFolderPath").text
tracks = root.findall(".//track")
train_vocals, test_vocals, train_mixes, test_mixes, train_accs, test_accs = list(), list(), list(), list(), list(), list()
for track in tracks:
# Get mix and vocal instruments
vocals = create_sample(db_path, track.xpath(".//instrument[instrumentName='Voice']")[0])
mix = create_sample(db_path, track.xpath(".//instrument[instrumentName='Mix']")[0])
[acc_path] = subtract_audio([mix.path], [vocals.path])
acc = Sample(acc_path, vocals.sample_rate, vocals.channels, vocals.duration) # Accompaniment has same signal properties as vocals and mix
if track.xpath("./databaseSplit")[0].text == "Training":
train_vocals.append(vocals)
train_mixes.append(mix)
train_accs.append(acc)
else:
test_vocals.append(vocals)
test_mixes.append(mix)
test_accs.append(acc)
return [train_mixes, train_accs, train_vocals], [test_mixes, test_accs, test_vocals]
def getCCMixter(xml_path):
tree = etree.parse(xml_path)
root = tree.getroot()
db_path = root.find("./databaseFolderPath").text
tracks = root.findall(".//track")
samples = list()
for track in tracks:
# Get mix and vocal instruments
voice = create_sample(db_path, track.xpath(".//instrument[instrumentName='Voice']")[0])
mix = create_sample(db_path, track.xpath(".//instrument[instrumentName='Mix']")[0])
acc = create_sample(db_path, track.xpath(".//instrument[instrumentName='Instrumental']")[0])
samples.append((mix, acc, voice))
return samples
def getIKala(xml_path):
tree = etree.parse(xml_path)
root = tree.getroot()
db_path = root.find("./databaseFolderPath").text
tracks = root.findall(".//track")
mixes, accs, vocals = list(), list(), list()
for track in tracks:
mix = create_sample(db_path, track.xpath(".//instrument[instrumentName='Mix']")[0])
orig_path = mix.path
mix_path = orig_path + "_mix.wav"
acc_path = orig_path + "_acc.wav"
voice_path = orig_path + "_voice.wav"
mix_audio, mix_sr = librosa.load(mix.path, sr=None, mono=False)
mix.path = mix_path
librosa.output.write_wav(mix_path, np.sum(mix_audio, axis=0), mix_sr)
librosa.output.write_wav(acc_path, mix_audio[0,:], mix_sr)
librosa.output.write_wav(voice_path, mix_audio[1, :], mix_sr)
voice = create_sample(mix.path, track.xpath(".//instrument[instrumentName='Voice']")[0])
voice.path = voice_path
acc = create_sample(mix.path, track.xpath(".//instrument[instrumentName='Instrumental']")[0])
acc.path = acc_path
mixes.append(mix)
accs.append(acc)
vocals.append(voice)
return [mixes, accs, vocals]
def getMedleyDB(xml_path):
tree = etree.parse(xml_path)
root = tree.getroot()
db_path = root.find("./databaseFolderPath").text
mixes, accs, vocals = list(), list(), list()
tracks = root.xpath(".//track")
for track in tracks:
instrument_paths = list()
# Mix together vocals, if they exist
vocal_tracks = track.xpath(".//instrument[instrumentName='Voice']/relativeFilepath") + \
track.xpath(".//instrument[instrumentName='Voice']/relativeFilepath") + \
track.xpath(".//instrument[instrumentName='Voice']/relativeFilepath")
if len(vocal_tracks) > 0: # If there are vocals, get their file paths and mix them together
vocal_track = Input.Input.add_audio([db_path + os.path.sep + f.text for f in vocal_tracks], "vocalmix")
instrument_paths.append(vocal_track)
vocals.append(Sample.from_path(vocal_track))
else: # Otherwise append duration of track so silent input can be generated later on-the-fly
duration = float(track.xpath("./instrumentList/instrument/length")[0].text)
vocals.append(duration)
# Mix together accompaniment, if it exists
acc_tracks = track.xpath(".//instrument[not(instrumentName='Voice') and not(instrumentName='Mix') and not(instrumentName='Instrumental')]/relativeFilepath") #TODO # We assume that there is no distinction between male/female here
if len(acc_tracks) > 0: # If there are vocals, get their file paths and mix them together
acc_track = Input.Input.add_audio([db_path + os.path.sep + f.text for f in acc_tracks], "accmix")
instrument_paths.append(acc_track)
accs.append(Sample.from_path(acc_track))
else: # Otherwise append duration of track so silent input can be generated later on-the-fly
duration = float(track.xpath("./instrumentList/instrument/length")[0].text)
accs.append(duration)
# Mix together vocals and accompaniment
mix_track = Input.Input.add_audio(instrument_paths, "totalmix")
mixes.append(Sample.from_path(mix_track))
return [mixes, accs, vocals]
def getFMA(xml_path):
tree = etree.parse(xml_path)
root = tree.getroot()
db_path = root.find("./databaseFolderPath").text
mixes, accs, vocals = list(), list(), list()
vocal_tracks = root.xpath(".//track/instrumentList/instrument[instrumentName='Mix']")
instrumental_tracks = root.xpath(".//track/instrumentList/instrument[instrumentName='Instrumental']")
for instr in vocal_tracks:
mixes.append(create_sample(db_path,instr))
for instr in instrumental_tracks:
mixes.append(create_sample(db_path,instr))
accs.append(create_sample(db_path,instr))
return mixes, accs, vocals