def gen(fname):
print(fname)
samples, samplingrate = loadAudio(fname)
lmfcc, mspec = mfcc(samples, liftering=False)
wordTrans = list(path2info(fname)[2])
phoneTrans = words2phones(wordTrans, prondict, addShortPause=True)
hmms = concatHMMs(phoneHMMs, phoneTrans)
stateTrans = [
phone + '_' + str(stateid) for phone in phoneTrans
for stateid in range(nstates[phone])
]
stateTrans_idx = list(map(stateList.index, stateTrans))
aligned = forcedAlignment(lmfcc, hmms, stateTrans_idx)
return aligned, lmfcc, mspec
import numpy as np
from lab1 import proto
import matplotlib.pyplot as plt
from lab1.tools import *
from scipy.cluster.hierarchy import linkage, dendrogram
import scipy
data = np.load('lab1_data.npz')['data']
example_data = np.load('lab1_example.npz')['example'].item()
target = 'lmfcc'
plt.figure(figsize=(10, 20))
ax = plt.subplot(8, 1, 1)
ax.set_title('Samples')
ax.set_yticklabels([])
ax.set_xticklabels([])
plt.plot(example_data['samples'], linewidth=0.5)
output = proto.mfcc(example_data['samples'])
plt.savefig('Results/part4.png', bbox_inches='tight')
plt.show()
import numpy as np
import matplotlib.pyplot as plt
from sklearn.mixture import GaussianMixture
from lab1 import proto
loaded = np.load('lab1_data.npz')
data = loaded['data']
size = 12
target = 'lmfcc'
giant_mfcc = []
seven = []
unrelated = []
i = 0
for d in data:
out = proto.mfcc(d['samples'])
giant_mfcc.append(out)
#voice seven
if i in [16, 17, 38, 39]:
seven.append(out)
if i in [5,35, 20]:
unrelated.append(out)
i += 1
giant_mfcc = np.vstack(giant_mfcc)
est = GaussianMixture(size)
est.fit(giant_mfcc)
import numpy as np
from lab1 import proto
import matplotlib.pyplot as plt
loaded = np.load('lab1_data.npz')
data = loaded['data']
target = 'lmfcc'
giant_mfcc = []
giant_mspec = []
for d in data:
out, mspec = proto.mfcc(d['samples'], liftering=False)
giant_mfcc.append(out)
giant_mspec.append(mspec)
giant_mfcc = np.vstack(giant_mfcc)
giant_mspec = np.vstack(giant_mspec)
# Answer for part 5
correlation_mfcc = np.corrcoef(giant_mfcc)
correlation_mspec = np.corrcoef(giant_mspec)
plt.figure(figsize=(20, 10))
ax = plt.subplot(1, 2, 1)
ax.set_title('MFCC')
ax.set_yticklabels([])
ax.set_xticklabels([])
plt.pcolormesh(correlation_mfcc)
ax = plt.subplot(1, 2, 2)
ax.set_title('MSPEC')
ax.set_yticklabels([])
#Get stateList
'''
#with open('stateList.pkl', 'wb') as f:
# pickle.dump(stateList, f)
'''
phoneHMMs = np.load('lab3/lab2_models.npz')['phoneHMMs'].item()
phones = sorted(phoneHMMs.keys())
nstates = {phone: phoneHMMs[phone]['means'].shape[0] for phone in phones}
stateList = [ph + '_' + str(id) for ph in phones for id in range(nstates[ph])]
# with open('lab3/stateList.pkl', 'rb') as f:
# stateList = pickle.load(f)
fname = 'lab3/asset/tidigits/disc_4.1.1/tidigits/train/man/nw/z43a.wav'
samples, samplingrate = loadAudio(fname)
lmfcc = mfcc(samples)
wordTrans = list(path2info(fname)[2])
phoneTrans = words2phones(wordTrans, prondict, addShortPause=True)
hmms = concatHMMs(phoneHMMs, phoneTrans)
stateTrans = [
phone + '_' + str(stateid) for phone in phoneTrans
for stateid in range(nstates[phone])
]
stateTrans_idx = list(map(stateList.index, stateTrans))
aligned = forcedAlignment(lmfcc, hmms, stateTrans)
frames2trans(aligned, outfilename='z43a.lab')
print("Done")