def test_interpolate_hmm(self):
acmodel1 = AcModel()
acmodel1.load_htk( os.path.join(HERE, "1-hmmdefs") )
acmodel2 = AcModel()
acmodel2.load_htk( os.path.join(HERE,"2-hmmdefs") )
ahmm1=acmodel1.get_hmm('a')
ahmm2=acmodel2.get_hmm('a')
# transitions
# (notice that the transition of 'a' in acmodel1 is in a macro.)
a1transition = [macro["transition"] for macro in acmodel1.macros if macro.get('transition',None)][0]
transitions = [ a1transition['definition'],ahmm2.definition['transition'] ]
trs = self.lin.linear_transitions( transitions, [1,0])
compare(trs,a1transition['definition'])
self.assertTrue(compare(trs,a1transition['definition']))
acmodel1.fill_hmms()
transitions = [ ahmm1.definition['transition'],ahmm2.definition['transition'] ]
trs = self.lin.linear_transitions( transitions, [1,0])
self.assertTrue(compare(trs,ahmm1.definition['transition']))
trs = self.lin.linear_transitions( transitions, [0,1])
self.assertTrue(compare(trs,ahmm2.definition['transition']))
# states
# (notice that the state 2 of 'a' in acmodel1 is in a macro.)
states = [ ahmm1.definition['states'],ahmm2.definition['states'] ]
sts = self.lin.linear_states( states, [1,0])
compare(sts,ahmm1.definition['states'],verbose=True)
self.assertTrue(compare(sts,ahmm1.definition['states']))
sts = self.lin.linear_states( states, [0,1])
self.assertTrue(compare(sts,ahmm2.definition['states']))
def test_monophones(self):
acmodel1 = AcModel()
acmodel1.load( os.path.join(MODEL_PATH,"models-fra") )
acmodel2 = acmodel1.extract_monophones()
acmodel2.save(os.path.join(HERE,'fra-mono'))
self.assertTrue( os.path.isfile( os.path.join(HERE, 'fra-mono','hmmdefs')) )
self.assertTrue( os.path.isfile( os.path.join(HERE, 'fra-mono','monophones.repl')) )
self.assertFalse( os.path.isfile( os.path.join(HERE, 'fra-mono','tiedlist')) )
os.remove( os.path.join(HERE, 'fra-mono','hmmdefs') )
os.remove( os.path.join(HERE, 'fra-mono','monophones.repl') )
os.rmdir( os.path.join(HERE,'fra-mono') )
self.assertEqual( len(acmodel2.hmms), 38 )
def test_merge(self):
acmodel1 = AcModel()
acmodel1.load_htk( os.path.join(HERE,"1-hmmdefs") )
acmodel2 = AcModel()
acmodel2.load_htk( os.path.join(HERE,"2-hmmdefs") )
(appended,interpolated,keeped,changed) = acmodel2.merge_model(acmodel1,gamma=0.5)
self.assertEqual(interpolated, 2) # acopy, a
self.assertEqual(appended, 1) # i
self.assertEqual(keeped, 1) # e
self.assertEqual(changed, 0)
self.__test_states( acmodel2.get_hmm('a').definition['states'] )
self.__test_transition( acmodel2.get_hmm('a').definition['transition'] )
def _test_load_save(self, acmodel):
# Save temporary the loaded model into a file
tmpfile = self.hmmdefs+".copy"
self.acmodel.save_htk( tmpfile )
# Load the temporary file into a new model
acmodelcopy = AcModel()
acmodelcopy.load_htk(tmpfile)
# Compare original and copy
self.assertEqual(len(self.acmodel.hmms),len(acmodelcopy.hmms))
for hmm,hmmcopy in zip(self.acmodel.hmms,acmodelcopy.hmms):
self.assertEqual(hmm.name,hmmcopy.name)
self.assertTrue(compare(hmm.definition,hmmcopy.definition))
self.assertTrue(compare(self.acmodel.macros,acmodelcopy.macros))
os.remove(tmpfile)
def testMix(self):
acmodel1 = AcModel()
hmm1 = HMM()
hmm1.create_proto( 25 )
hmm1.name = "y"
acmodel1.append_hmm( hmm1 )
acmodel1.repllist.add("y","j")
acmodel2 = AcModel()
hmm2 = HMM()
hmm2.create_proto( 25 )
hmm2.name = "j"
hmm3 = HMM()
hmm3.create_proto( 25 )
hmm3.name = "y"
acmodel2.hmms.append( hmm2 )
acmodel2.hmms.append( hmm3 )
acmodel2.repllist.add("y","y")
acmodel2.repllist.add("j","j")
modelmixer = ModelMixer()
modelmixer.set_models( acmodel1,acmodel2 )
outputdir = os.path.join(MODELDIR, "models-test")
modelmixer.mix( outputdir, gamma=1. )
mixedh1 = AcModel()
mixedh1.load( outputdir )
shutil.rmtree( outputdir )
def test_fill(self):
acmodel1 = AcModel()
acmodel1.load_htk( os.path.join(HERE,"1-hmmdefs") )
ahmm1=acmodel1.get_hmm('a')
a1transition = [macro["transition"] for macro in acmodel1.macros if macro.get('transition',None)][0]
acmodel1.fill_hmms()
self.assertTrue(compare(ahmm1.definition['transition'],a1transition['definition']))
def testMixData(self):
modelmixer = ModelMixer()
modelmixer.load(self._modelL2dir, self._modelL1dir )
outputdir = os.path.join(MODELDIR, "models-eng-fra")
modelmixer.mix( outputdir, gamma=0.5 )
self.assertTrue( os.path.exists( outputdir ) )
acmodel1 = AcModel()
acmodel1.load_htk( os.path.join(self._modelL2dir, "hmmdefs") )
acmodel1 = acmodel1.extract_monophones()
acmodel2 = AcModel()
acmodel2.load_htk( os.path.join(os.path.join(MODELDIR, "models-eng-fra"), "hmmdefs") )
shutil.rmtree( outputdir )
def test_initializer_without_corpus(self):
corpus = TrainingCorpus()
workdir = os.path.join(HERE,"working")
os.mkdir( workdir )
shutil.copy( os.path.join(HERE, "protos","vFloors"), workdir )
initial = HTKModelInitializer( corpus, workdir )
corpus.datatrainer.protodir = os.path.join(HERE, "protos")
initial.create_model()
hmm1 = HMM()
hmm2 = HMM()
hmm1.load( os.path.join( workdir, "@@.hmm") )
hmm2.load( os.path.join(HERE, "protos", "@@.hmm") )
self.assertTrue(compare(hmm1.definition,hmm2.definition))
hmm1.load( os.path.join( workdir, "sil.hmm") )
hmm2.load( os.path.join(HERE, "protos", "sil.hmm") )
corpus.datatrainer.fix_proto(protofilename=os.path.join(HERE, "protos", "proto.hmm"))
hmm2.load( os.path.join(HERE, "protos", "proto.hmm") )
hmm1.load( os.path.join(workdir, "gb.hmm") )
self.assertTrue(compare(hmm1.definition,hmm2.definition))
hmm1.load( os.path.join(workdir, "dummy.hmm") )
self.assertTrue(compare(hmm1.definition,hmm2.definition))
acmodel = AcModel()
acmodel.load_htk( os.path.join( workdir,"hmmdefs") )
shutil.rmtree( workdir )
os.remove( os.path.join(HERE, "protos", "proto.hmm") )
def load(self, modelTextDir, modelSpkDir):
"""
Load the acoustic models from their directories.
@param modelTextDir (str)
@param modelSpkDir (str)
"""
modelText = AcModel()
modelSpk = AcModel()
# Load the acoustic models.
modelText.load( modelTextDir )
modelSpk.load( modelSpkDir )
self.set_models(modelText, modelSpk)
args = parser.parse_args()
# ----------------------------------------------------------------------------
if not os.path.isdir( args.o ):
print "Error:",args.o,"must be an existing directory."
sys.exit(1)
if not os.path.isfile( args.i ):
print "Error:",args.i,"must be an acoutic model file (HTK-ASCII format)."
sys.exit(1)
if not args.quiet is True:
print "Loading AC:",
acmodel1 = AcModel()
acmodel1.load_htk( args.i )
if not args.quiet is True:
print "... done"
# ----------------------------------------------------------------------------
acmodel = acmodel1.extract_monophones()
for hmm in acmodel.hmms:
filename = os.path.join( args.o, hmm.name )
filename = filename + ".hmm"
if not args.quiet is True:
print hmm.name,filename
hmm.save( filename )
def test_replace_phones(self):
acmodel1 = AcModel()
acmodel1.load( os.path.join(MODEL_PATH,"models-fra") )
acmodel1.replace_phones( reverse=False )
acmodel1.replace_phones( reverse=True )
acmodel2 = AcModel()
acmodel2.load( os.path.join(MODEL_PATH,"models-fra") )
for h1 in acmodel1.hmms:
h2 = acmodel2.get_hmm( h1.name )
self.assertTrue(compare(h1.definition['transition'],h2.definition['transition']))
self.assertTrue(compare(h1.definition['states'],h2.definition['states']))
def test_no_merge(self):
nbhmms = len(self.acmodel.hmms)
# Try to merge with the same model!
acmodel2 = AcModel()
acmodel2.load_htk(self.hmmdefs)
(appended,interpolated,keeped,changed) = acmodel2.merge_model(self.acmodel,gamma=1.)
self.assertEqual(interpolated, 0)
self.assertEqual(appended, 0)
self.assertEqual(keeped, nbhmms)
self.assertEqual(changed, 0)
(appended,interpolated,keeped,changed) = acmodel2.merge_model(self.acmodel,gamma=0.5)
self.assertEqual(interpolated, nbhmms)
self.assertEqual(appended, 0)
self.assertEqual(keeped, 0)
self.assertEqual(changed, 0)
(appended,interpolated,keeped,changed) = acmodel2.merge_model(self.acmodel,gamma=0.)
self.assertEqual(interpolated, 0)
self.assertEqual(appended, 0)
self.assertEqual(keeped, 0)
self.assertEqual(changed, nbhmms)
# Try to merge with a different MFCC parameter kind model...
acmodel2 = AcModel()
acmodel2.load_htk( os.path.join(MODEL_PATH,"models-cat","hmmdefs") )
with self.assertRaises(TypeError):
acmodel2.merge_model(self.acmodel,gamma=1.)
def setUp(self):
self.hmmdefs = os.path.join(MODEL_PATH,"models-jpn","hmmdefs")
self.acmodel = AcModel()
self.acmodel.load_htk( self.hmmdefs )
class TestAcModel(unittest.TestCase):
#This one takes too much time to be tested each time....
# def test_load_all_models(self):
# models = glob.glob(os.path.join(MODEL_PATH,"models-*","hmmdefs"))
# for hmmdefs in models:
# acmodel = AcModel( hmmdefs )
# self._test_load_save( acmodel )
def setUp(self):
self.hmmdefs = os.path.join(MODEL_PATH,"models-jpn","hmmdefs")
self.acmodel = AcModel()
self.acmodel.load_htk( self.hmmdefs )
def test_load_save(self):
self._test_load_save( self.hmmdefs )
def _test_load_save(self, acmodel):
# Save temporary the loaded model into a file
tmpfile = self.hmmdefs+".copy"
self.acmodel.save_htk( tmpfile )
# Load the temporary file into a new model
acmodelcopy = AcModel()
acmodelcopy.load_htk(tmpfile)
# Compare original and copy
self.assertEqual(len(self.acmodel.hmms),len(acmodelcopy.hmms))
for hmm,hmmcopy in zip(self.acmodel.hmms,acmodelcopy.hmms):
self.assertEqual(hmm.name,hmmcopy.name)
self.assertTrue(compare(hmm.definition,hmmcopy.definition))
self.assertTrue(compare(self.acmodel.macros,acmodelcopy.macros))
os.remove(tmpfile)
def test_get_hmm(self):
with self.assertRaises(ValueError):
self.acmodel.get_hmm('Q')
Nhmm = self.acmodel.get_hmm('N')
self.__test_states( Nhmm.definition['states'] )
self.__test_transition( Nhmm.definition['transition'] )
def test_append_hmm(self):
with self.assertRaises(TypeError):
self.acmodel.append_hmm({'toto':None})
hmm = HMM()
with self.assertRaises(TypeError):
self.acmodel.append_hmm(hmm)
Nhmm = self.acmodel.get_hmm('N')
with self.assertRaises(ValueError):
self.acmodel.append_hmm(Nhmm)
Newhmm = copy.deepcopy(Nhmm)
Newhmm.name = "NewN"
self.acmodel.append_hmm(Newhmm)
def test_pop_hmm(self):
self.acmodel.pop_hmm("N")
with self.assertRaises(ValueError):
self.acmodel.get_hmm( "N" )
def test_load_hmm(self):
hmm = HMM()
hmm.load( os.path.join(HERE,"N-hmm") )
self.__test_states( hmm.definition['states'] )
self.__test_transition( hmm.definition['transition'] )
def test_save_hmm(self):
hmm = HMM()
hmm.load( os.path.join(HERE,"N-hmm") )
hmm.save(os.path.join(HERE,"N-hmm-copy"))
newhmm = HMM()
newhmm.load(os.path.join(HERE,"N-hmm-copy"))
os.remove(os.path.join(HERE,'N-hmm-copy'))
self.assertEqual(hmm.name,newhmm.name)
self.assertTrue(compare(hmm.definition,newhmm.definition))
def test_fill(self):
acmodel1 = AcModel()
acmodel1.load_htk( os.path.join(HERE,"1-hmmdefs") )
ahmm1=acmodel1.get_hmm('a')
a1transition = [macro["transition"] for macro in acmodel1.macros if macro.get('transition',None)][0]
acmodel1.fill_hmms()
self.assertTrue(compare(ahmm1.definition['transition'],a1transition['definition']))
def test_no_merge(self):
nbhmms = len(self.acmodel.hmms)
# Try to merge with the same model!
acmodel2 = AcModel()
acmodel2.load_htk(self.hmmdefs)
(appended,interpolated,keeped,changed) = acmodel2.merge_model(self.acmodel,gamma=1.)
self.assertEqual(interpolated, 0)
self.assertEqual(appended, 0)
self.assertEqual(keeped, nbhmms)
self.assertEqual(changed, 0)
(appended,interpolated,keeped,changed) = acmodel2.merge_model(self.acmodel,gamma=0.5)
self.assertEqual(interpolated, nbhmms)
self.assertEqual(appended, 0)
self.assertEqual(keeped, 0)
self.assertEqual(changed, 0)
(appended,interpolated,keeped,changed) = acmodel2.merge_model(self.acmodel,gamma=0.)
self.assertEqual(interpolated, 0)
self.assertEqual(appended, 0)
self.assertEqual(keeped, 0)
self.assertEqual(changed, nbhmms)
# Try to merge with a different MFCC parameter kind model...
acmodel2 = AcModel()
acmodel2.load_htk( os.path.join(MODEL_PATH,"models-cat","hmmdefs") )
with self.assertRaises(TypeError):
acmodel2.merge_model(self.acmodel,gamma=1.)
def test_merge(self):
acmodel1 = AcModel()
acmodel1.load_htk( os.path.join(HERE,"1-hmmdefs") )
acmodel2 = AcModel()
acmodel2.load_htk( os.path.join(HERE,"2-hmmdefs") )
(appended,interpolated,keeped,changed) = acmodel2.merge_model(acmodel1,gamma=0.5)
self.assertEqual(interpolated, 2) # acopy, a
self.assertEqual(appended, 1) # i
self.assertEqual(keeped, 1) # e
self.assertEqual(changed, 0)
self.__test_states( acmodel2.get_hmm('a').definition['states'] )
self.__test_transition( acmodel2.get_hmm('a').definition['transition'] )
def test_replace_phones(self):
acmodel1 = AcModel()
acmodel1.load( os.path.join(MODEL_PATH,"models-fra") )
acmodel1.replace_phones( reverse=False )
acmodel1.replace_phones( reverse=True )
acmodel2 = AcModel()
acmodel2.load( os.path.join(MODEL_PATH,"models-fra") )
for h1 in acmodel1.hmms:
h2 = acmodel2.get_hmm( h1.name )
self.assertTrue(compare(h1.definition['transition'],h2.definition['transition']))
self.assertTrue(compare(h1.definition['states'],h2.definition['states']))
def test_monophones(self):
acmodel1 = AcModel()
acmodel1.load( os.path.join(MODEL_PATH,"models-fra") )
acmodel2 = acmodel1.extract_monophones()
acmodel2.save(os.path.join(HERE,'fra-mono'))
self.assertTrue( os.path.isfile( os.path.join(HERE, 'fra-mono','hmmdefs')) )
self.assertTrue( os.path.isfile( os.path.join(HERE, 'fra-mono','monophones.repl')) )
self.assertFalse( os.path.isfile( os.path.join(HERE, 'fra-mono','tiedlist')) )
os.remove( os.path.join(HERE, 'fra-mono','hmmdefs') )
os.remove( os.path.join(HERE, 'fra-mono','monophones.repl') )
os.rmdir( os.path.join(HERE,'fra-mono') )
self.assertEqual( len(acmodel2.hmms), 38 )
def test_proto(self):
h1 = HtkIO()
h1.write_hmm_proto( 25, os.path.join(HERE,"proto_from_htkio") )
h2 = HMM()
h2.create_proto( 25 )
h2.save( os.path.join(HERE,"proto_from_hmm") )
m1 = HMM()
m1.load( os.path.join(HERE,"proto_from_htkio") )
m2 = HMM()
m2.load( os.path.join(HERE,"proto_from_hmm") )
self.assertTrue(compare(m1.definition['transition'],m2.definition['transition']))
self.assertTrue(compare(m1.definition['states'],m2.definition['states']))
os.remove( os.path.join(HERE,"proto_from_hmm") )
os.remove( os.path.join(HERE,"proto_from_htkio") )
def __test_transition(self, transition):
self.assertEqual(transition['dim'], 5)
matrix = transition['matrix']
for i in range(len(matrix)-1):
# the last vector is always 0.!
vector=matrix[i]
self.assertEqual(1.0, round(sum(vector),4))
def __test_states(self, states):
for item in states: # a dict
state = item['state']
streams = state['streams']
for s in streams: # a list
mixtures = s['mixtures']
for mixture in mixtures: # a list of dict
#self.assertEqual(type(mixture['weight']),float)
pdf = mixture['pdf']
self.assertEqual(pdf['mean']['dim'], 25)
self.assertEqual(len(pdf['mean']['vector']),25)
self.assertEqual(pdf['covariance']['variance']['dim'], 25)
self.assertEqual(len(pdf['covariance']['variance']['vector']), 25)
self.assertEqual(type(pdf['gconst']),float)
mxg = parser.add_mutually_exclusive_group(required=True)
mxg.add_argument("-o", metavar="file", required=False, help='Output file name')
mxg.add_argument("-O", metavar="file", required=False, help='Output directory name')
if len(sys.argv) <= 1:
sys.argv.append('-h')
args = parser.parse_args()
# ----------------------------------------------------------------------------
if not args.quiet is True:
print "Loading AC 1:",
acmodel1 = AcModel()
if os.path.isfile( args.i ):
acmodel1.load_htk( args.i )
else:
acmodel1.load( args.i )
if not args.quiet is True:
print "... done"
if not args.quiet is True:
print "Loading AC 2:",
acmodel2 = AcModel()
if os.path.isfile( args.I ):
acmodel2.load_htk( args.I )
else:
acmodel2.load( args.I )
if not args.quiet is True:
