re-aligned mlf file include less files than original mlf file. Therefore the scp file should also be updated accordingly, when re-estimation is performed. this bug is fixed.

acoustic_model/fame_hmm.py

@@ -27,8 +27,8 @@ extract_features  = 0
 flat_start		  = 0
 train_model_without_sp = 0
 add_sp = 0
-train_model_with_re_aligned_mlf = 1
-train_triphone = 0
+train_model_with_re_aligned_mlf = 0
+train_triphone = 1
 
 
 
@@ -45,9 +45,6 @@ lexicon_oov = os.path.join(default.fame_dir, 'lexicon', 'lex.oov')
 
 config_dir = os.path.join(default.htk_dir, 'config')
 
-sil_hed		 = os.path.join(config_dir, 'sil.hed')
-prototype    = os.path.join(config_dir, proto_name)
-
 model_dir     = os.path.join(default.htk_dir, 'model')
 model0_dir    = os.path.join(model_dir, 'hmm0')
 model1_dir    = os.path.join(model_dir, 'hmm1')
@@ -72,6 +69,7 @@ fh.make_new_directory(label_dir, existing_dir='leave')
 hcompv_scp_train = os.path.join(tmp_dir, 'train.scp')
 mlf_file_train   = os.path.join(label_dir, 'train_phone.mlf')
 mlf_file_train_aligned = os.path.join(label_dir, 'train_phone_aligned.mlf')
+hcompv_scp_train_updated = hcompv_scp_train.replace('.scp', '_updated.scp')
 
 ## testing
 htk_stimmen_dir = os.path.join(default.htk_dir, 'stimmen')
@@ -304,6 +302,13 @@ if train_model_with_re_aligned_mlf:
 		mlf_file_train_aligned, 		
 		os.path.join(label_dir,	'train_word.mlf'), 
 		hcompv_scp_train)
+
+	print('>>> updating the script file... ')
+	chtk.update_script_file(
+		mlf_file_train_aligned, 
+		mlf_file_train, 
+		hcompv_scp_train, 
+		hcompv_scp_train_updated)
 	print("elapsed time: {}".format(time.time() - timer_start))
 
 	print('>>> re-estimation... ')
@@ -314,11 +319,11 @@ if train_model_with_re_aligned_mlf:
 		model1sp2_dir, 
 		os.path.join(model1sp_dir, 'iter'+str(niter)), 
 		improvement_threshold, 
-		hcompv_scp_train, 
+		hcompv_scp_train_updated, 
 		os.path.join(htk_stimmen_dir, 'mfc'), 
 		'mfc', 
 		os.path.join(htk_stimmen_dir, 'word_lattice.ltc'), 
-		mlf_file=mlf_file_train, 
+		mlf_file=mlf_file_train_aligned, 
 		lexicon_file=os.path.join(htk_stimmen_dir, 'lexicon_recognition.dic'), 
 		model_type='monophone_with_sp'
 		)
@@ -327,19 +332,25 @@ if train_model_with_re_aligned_mlf:
 
 ## ======================= train triphone =======================
 if train_triphone:
-	triphone_mlf = os.path.join(default.htk_dir, 'label', 'train_triphone.mlf')
-	macros = os.path.join(model_dir, 'hmm1_tri', 'iter0', 'macros')
-	hmmdefs = os.path.join(model_dir, 'hmm1_tri', 'iter0', 'hmmdefs')
 	model_out_dir = os.path.join(model_dir, 'hmm1_tri', 'iter1')
-	run_command([
-		'HERest', '-B', 
-		'-C', config_train,
-		'-I', triphone_mlf,
-		'-t', '250.0', '150.0', '1000.0',
-		'-s', 'stats' 
-		'-S', hcompv_scp_train,
-		'-H', macros,
-		'-H', hmmdefs,
-		'-M', model_out_dir, 
-		os.path.join(config_dir, 'triphonelist.txt')
-	])
+
+	triphonelist_txt = os.path.join(config_dir, 'triphonelist_txt')
+	triphone_mlf = os.path.join(default.htk_dir, 'label', 'train_triphone.mlf')
+
+	chtk.make_triphonelist(
+		triphonelist_txt, 
+		triphone_mlf, 
+		mlf_file_train_aligned)
+
+	#run_command([
+	#	'HERest', '-B', 
+	#	'-C', config_train,
+	#	'-I', triphone_mlf,
+	#	'-t', '250.0', '150.0', '1000.0',
+	#	'-s', 'stats' 
+	#	'-S', hcompv_scp_train,
+	#	'-H', macros,
+	#	'-H', hmmdefs,
+	#	'-M', model_out_dir, 
+	#	os.path.join(config_dir, 'triphonelist.txt')
+	#])