def populate(self, directory):
'''
Analyzes all audio files in a directory and add their fingerprint
to the database. Writes result to disk when done.
Args:
directory: directory containing audio files.
'''
audio_files = sorted(soundfiles.find_files(directory + '/*'))
c.notice("Reading {} files...".format(len(audio_files)))
i = 0
for filename in audio_files:
i += 1
c.write('\r\t-file ' + str(i) + ' of ' + str(len(audio_files)))
signal = soundfiles.load_signal(filename)
self.add(fingerprint.get_tokens(signal))
c.write("\n")
c.notice("Writing database to disk...")
self.save()
c.succes("Database written")
def populate(self, directory):
'''
Analyzes all audio files in a directory and add their fingerprint
to the database. Writes result to disk when done.
Args:
directory: directory containing audio files.
'''
audio_files = sorted(soundfiles.find_files(directory + '/*'))
c.notice("Reading {} files...".format(len(audio_files)))
i = 0
for filename in audio_files:
i += 1
c.write('\r\t-file ' + str(i) + ' of ' + str(len(audio_files)))
signal = soundfiles.load_signal(filename)
self.add(fingerprint.get_tokens(signal))
c.write("\n")
c.notice("Writing database to disk...")
self.save()
c.succes("Database written")
exit()
if args.train:
database.populate(args.train)
if args.f:
if not database.read:
c.notice('Loading database from disk...')
database.load()
c.succes('Loaded {} entries.'.format(database.get_size()))
c.notice('Converting database to classifier...')
classifier = database.as_classifier()
c.notice('Loading \'{}\' from disk...'.format(args.f))
signal = load_signal(args.f)
if signal is None:
c.fatal('This file could not be loaded')
parser.print_help()
exit()
c.notice('Analyzing \'{}\'...'.format(signal.get_filename()))
tokens = get_tokens(signal)
c.notice("Classifying...")
match = classifier.classify(tokens)
if match:
c.notice("File \'{}\' matches with database entry {}".format(
signal.filename, match))
else:
exit()
if args.train:
database.populate(args.train)
if args.f:
if not database.read:
c.notice('Loading database from disk...')
database.load()
c.succes('Loaded {} entries.'.format(database.get_size()))
c.notice('Converting database to classifier...')
classifier = database.as_classifier()
c.notice('Loading \'{}\' from disk...'.format(args.f))
signal = load_signal(args.f)
if signal is None:
c.fatal('This file could not be loaded')
parser.print_help()
exit()
c.notice('Analyzing \'{}\'...'.format(signal.get_filename()))
tokens = get_tokens(signal)
c.notice("Classifying...")
match = classifier.classify(tokens)
if match:
c.notice("File \'{}\' matches with database entry {}".format(
signal.filename, match
))
import soundfiles
import random
import matplotlib.pyplot as plt
import wave
import numpy as np
import os
if not os.path.exists("audio/pokemon"):
os.makedirs("audio/pokemon")
directory = "training/pokemon/*.wav"
files = soundfiles.find_files(directory)
for f in files:
signal = soundfiles.load_signal(f)
'''
initialize the suffix of the new file and the new samples
'''
suffix = ""
new_samples = signal.samples
if random.random() < 0.5:
'''
add noise
'''
new_samples += 0.05 - 0.1 * np.random.rand(len(new_samples))
suffix += "-noise"
if random.random() < 0.5:
import soundfiles
import random
import matplotlib.pyplot as plt
import wave
import numpy as np
import os
if not os.path.exists("audio/pokemon"):
os.makedirs("audio/pokemon")
directory = "training/pokemon/*.wav"
files = soundfiles.find_files(directory)
for f in files:
signal = soundfiles.load_signal(f)
'''
initialize the suffix of the new file and the new samples
'''
suffix = ""
new_samples = signal.samples
if random.random() < 0.5:
'''
add noise
'''
new_samples += 0.05 - 0.1 * np.random.rand(len(new_samples))
suffix += "-noise"
if random.random() < 0.5:
'''
pokemon_nr = "103" # "003"
filenames = [
"training/pokemon/{}.wav".format(pokemon_nr),
"audio/presentatie/{}-divided.wav".format(pokemon_nr),
"audio/presentatie/{}-noise.wav".format(pokemon_nr),
"audio/presentatie/{}-translated.wav".format(pokemon_nr),
]
n = len(filenames)
fig, ax = plt.subplots(n)
for i in xrange(n):
filename = filenames[i]
signal = soundfiles.load_signal(filename)
fingerprints = fingerprint.get_fingerprints(signal, 1024, 2)
x = [time for (time, peaks) in fingerprints for freq in peaks]
y = [freq for (time, peaks) in fingerprints for freq in peaks]
if "translated" in filename:
shift = 16384 / 1024
x = [e - shift for e in x]
for h in xrange(128, 256, 32):
ax[i].axhline(y=h, color="grey", alpha=0.25)
plot = ax[i].scatter(x, y, color="red")
ax[i].set_xlim((0, 23))
ax[i].set_ylim((128, 256))
print "Start testing classifier with {} files\n".format(len(filenames))
shuffle(filenames)
correct_match = 0
wrong_match = 0
no_match = 0
wrong_matches = ""
correct_matches = ""
no_matches = ""
for filename in filenames:
print "Classifying {}".format(filename)
signal = load_signal(filename)
tokens = get_tokens(signal)
match = classifier.classify(tokens)
if match is None:
print "-> No match found"
no_matches += filename.split("/")[2][:-4]
no_match += 1
elif match.split("/")[2][:3] == filename.split("/")[2][:3]:
print "-> Correct match found"
correct_match += 1
correct_matches += filename.split("/")[2][:-4]
else:
print "-> Wrong match found"
wrong_matches += filename.split("/")[2][:-4]
wrong_match += 1
print "Start testing classifier with {} files\n".format(len(filenames))
shuffle(filenames)
correct_match = 0
wrong_match = 0
no_match = 0
wrong_matches = ""
correct_matches = ""
no_matches = ""
for filename in filenames:
print "Classifying {}".format(filename)
signal = load_signal(filename)
tokens = get_tokens(signal)
match = classifier.classify(tokens)
if match is None:
print "-> No match found"
no_matches += filename.split("/")[2][:-4]
no_match += 1
elif match.split("/")[2][:3] == filename.split("/")[2][:3]:
print "-> Correct match found"
correct_match += 1
correct_matches += filename.split("/")[2][:-4]
else:
print "-> Wrong match found"
wrong_matches += filename.split("/")[2][:-4]
wrong_match += 1
pokemon_nr = "103" # "003"
filenames = [
"training/pokemon/{}.wav".format(pokemon_nr),
"audio/presentatie/{}-divided.wav".format(pokemon_nr),
"audio/presentatie/{}-noise.wav".format(pokemon_nr),
"audio/presentatie/{}-translated.wav".format(pokemon_nr)
]
n = len(filenames)
fig, ax = plt.subplots(n)
for i in xrange(n):
filename = filenames[i]
signal = soundfiles.load_signal(filename)
fingerprints = fingerprint.get_fingerprints(signal, 1024, 2)
x = [time for (time, peaks) in fingerprints for freq in peaks]
y = [freq for (time, peaks) in fingerprints for freq in peaks]
if "translated" in filename:
shift = 16384 / 1024
x = [e - shift for e in x]
for h in xrange(128, 256, 32):
ax[i].axhline(y=h, color="grey", alpha=0.25)
plot = ax[i].scatter(x, y, color="red")
ax[i].set_xlim((0, 23))
ax[i].set_ylim((128, 256))
