def adaptation(args):
if args.feat_type == 'mfcc':
datasetlist = ["energy", "cep", "vad"]
mask = "[0-12]"
features_folder = '/home/zeng/zeng/aishell/af2019-sr-devset-20190312/feature'
if args.feat_type == 'fb':
datasetlist = ["fb", "vad"]
mask = None
features_folder = '/home/zeng/zeng/aishell/af2019-sr-devset-20190312/feature'
# create feature server for loading feature from disk
feature_server = sidekit.FeaturesServer(
features_extractor=None,
feature_filename_structure=features_folder + "/{}.h5",
sources=None,
dataset_list=datasetlist,
mask=mask,
feat_norm="cmvn",
global_cmvn=None,
dct_pca=False,
dct_pca_config=None,
sdc=False,
sdc_config=None,
delta=True if args.delta else False,
double_delta=True if args.delta else False,
delta_filter=None,
context=None,
traps_dct_nb=None,
rasta=True,
keep_all_features=False)
enroll_idmap = sidekit.IdMap(os.getcwd() + '/task/idmap.h5')
ndx = sidekit.Ndx(os.getcwd() + '/task/dev_ndx.h5')
ubm = sidekit.Mixture()
ubm.read(os.getcwd() + '/model/ubm.h5')
enroll_stat = sidekit.StatServer(enroll_idmap,
distrib_nb=ubm.distrib_nb(),
feature_size=ubm.dim())
enroll_stat.accumulate_stat(ubm=ubm,
feature_server=feature_server,
seg_indices=range(enroll_stat.segset.shape[0]),
num_thread=args.num_thread)
enroll_stat.write(os.getcwd() + '/task/enroll_stat.h5')
print('MAP adaptation', end='')
regulation_factor = 16
enroll_sv = enroll_stat.adapt_mean_map_multisession(ubm, regulation_factor)
enroll_sv.write(os.getcwd() + '/task/enroll_sv.h5')
print('\rMAP adaptation done')
print('Compute scores', end='')
score = sidekit.gmm_scoring(ubm,
enroll_sv,
ndx,
feature_server,
num_thread=args.num_thread)
score.write(os.getcwd() + '/task/dev_score.h5')
print('\rCompute scores done')
def evaluate(self, explain=True):
"""
This method is used to evaluate the test set. It does so by"
- read the test_ndx file that contains the test set
- read the trained UBM model, and trained parameters (enroll_stat file)
- evaluate the test set using gmm_scoring and write the scores
- if explain=True, write the scores in a more readible way
Args:
explain (boolean): If True, write another text file that contain
the same information as the one within ubm_scores file but in a
readible way.
"""
############################# READING ############################
# Create Feature server
server = self.createFeatureServer()
# Read the index for the test datas
test_ndx = sidekit.Ndx.read(
os.path.join(self.BASE_DIR, "task", "test_ndx.h5"))
# Read the UBM model
ubm = sidekit.Mixture()
model_name = "ubm_{}.h5".format(self.NUM_GAUSSIANS)
ubm.read(os.path.join(self.BASE_DIR, "ubm", model_name))
filename = "enroll_stat_{}.h5".format(self.NUM_GAUSSIANS)
enroll_stat = sidekit.StatServer.read(
os.path.join(self.BASE_DIR, "stat", filename))
# MAP adaptation of enrollment speaker models
enroll_sv = enroll_stat.adapt_mean_map_multisession(
ubm=ubm,
r=3 # MAP regulation factor
)
############################ Evaluating ###########################
# Compute scores
scores_gmm_ubm = sidekit.gmm_scoring(ubm=ubm,
enroll=enroll_sv,
ndx=test_ndx,
feature_server=server,
num_thread=self.NUM_THREADS)
# Save the model's Score object
filename = "ubm_scores_{}.h5".format(self.NUM_GAUSSIANS)
scores_gmm_ubm.write(os.path.join(self.BASE_DIR, "result", filename))
# Explain the Analysis by writing more readible text file
if explain:
filename = "ubm_scores_{}_explained.txt".format(self.NUM_GAUSSIANS)
fout = open(os.path.join(self.BASE_DIR, "result", filename), "a")
fout.truncate(0) #clear content
modelset = list(scores_gmm_ubm.modelset)
segset = list(scores_gmm_ubm.segset)
scores = np.array(scores_gmm_ubm.scoremat)
for seg_idx, seg in enumerate(segset):
fout.write("Wav: {}\n".format(seg))
for speaker_idx, speaker in enumerate(modelset):
fout.write("\tSpeaker {}:\t{}\n"\
.format(speaker, scores[speaker_idx, seg_idx]))
fout.write("\n")
fout.close()
def evaluate(self, explain=True):
############################# READING ############################
# Create Feature server
server = self.createFeatureServer()
# Read the index for the test datas
test_ndx = sidekit.Ndx.read(
os.path.join(self.BASE_DIR, "task", "test_ndx.h5"))
# Read the UBM model
ubm = sidekit.Mixture()
model_name = "ubm_{}.h5".format(self.NUM_GUASSIANS)
ubm.read(os.path.join(self.BASE_DIR, "ubm", model_name))
############################ Evaluating ###########################
filename = "enroll_stat_{}.h5".format(self.NUM_GUASSIANS)
enroll_stat = sidekit.StatServer.read(
os.path.join(self.BASE_DIR, "ubm", filename))
# MAP adaptation of enrollment speaker models
enroll_sv = enroll_stat.adapt_mean_map_multisession(
ubm=ubm,
r=3 # MAP regulation factor
)
# Compute scores
scores_gmm_ubm = sidekit.gmm_scoring(ubm=ubm,
enroll=enroll_sv,
ndx=test_ndx,
feature_server=server,
num_thread=self.NUM_THREADS)
# Save the model's Score object
filename = "ubm_scores_{}.h5".format(self.NUM_GUASSIANS)
scores_gmm_ubm.write(os.path.join(self.BASE_DIR, "result", filename))
#write Analysis
if explain:
filename = "ubm_scores_explained_{}.txt".format(self.NUM_GUASSIANS)
fout = open(os.path.join(self.BASE_DIR, "result", filename), "a")
fout.truncate(0) #clear content
modelset = list(scores_gmm_ubm.modelset)
segset = list(scores_gmm_ubm.segset)
scores = np.array(scores_gmm_ubm.scoremat)
for seg_idx, seg in enumerate(segset):
fout.write("Wav: {}\n".format(seg))
for speaker_idx, speaker in enumerate(modelset):
fout.write("\tSpeaker {}:\t{}\n".format(
speaker, scores[speaker_idx, seg_idx]))
fout.write("\n")
fout.close()
num_thread=nbThread)
enroll_stat.write('data/stat_rsr2015_male_enroll.h5')
# Adapt the GMM speaker models from the UBM via a MAP adaptation
print('MAP adaptation of the speaker models')
regulation_factor = 3 # MAP regulation factor
enroll_sv = enroll_stat.adapt_mean_map_multisession(ubm, regulation_factor)
enroll_sv.write('data/sv_rsr2015_male_enroll.h5')
# Compute all trials and save scores in HDF5 format
print('Compute trial scores')
scores_gmm_ubm = sidekit.gmm_scoring(ubm,
enroll_sv,
test_ndx,
features_server,
num_thread=nbThread)
scores_gmm_ubm.write('scores/scores_gmm-ubm_rsr2015_male.h5')
# Plot DET curve and compute minDCF and EER
print('Plot the DET curve')
# Set the prior following NIST-SRE 2008 settings
prior = sidekit.logit_effective_prior(0.01, 10, 1)
# Initialize the DET plot to 2008 settings
dp = sidekit.DetPlot(window_style='sre10', plot_title='GMM-UBM_RSR2015_male')
dp.set_system_from_scores(scores_gmm_ubm, key, sys_name='GMM-UBM')
dp.create_figure()
dp.plot_rocch_det(0)
