THIS ERROR DOCUMENT IS FOR DEVELOPER USING MFA IN LINUX,UNIX:
FOLLOW ALL THE INSTRUCTION MENTIONED ON MFA WEBSITE.
ERROR NUMBER 1:
AS MENTIONED ON MFA WEBSITE YOU SHOULD USE PYTHON VERSION 3.X AND PIP
THEN USE COMMAND “python3” IN EVERY PROCESS MENTIONED IN BUILDING
FROM SOURCE PART.
ERROR NUMBER 2:
DESCRIPTION :
no such file ‘fstcompile’
POSSIBILITY 1:
YOU MAY NOT HAVE COMPILED KALDI.FOLLOW INSTRUCTION IN TOOLS AND SRC INSTALL FILES.
POSSIBILITY 2:
USE LATEST VERSION OF MFA “fstcompile” FILE IS LOCATED IN
“lib/thirdparty/bin”.
ERROR NUMBER 3:
DESCRIPTION :
Exception: The language ‘/home/kranti/montreal-forced-aligner-
master/pretrained_model/english’ is not currently included in the distribution, please
align via training or specify one of the following language names: english.
kranti@kranti:~/Montreal-Forced-Aligner-master$ bin/mfa_align -t ~/output -v ~/MFATestWav ~/SpanishDict/Spanishdict.txt ~/SpanishModel spanish
Traceback (most recent call last):
File "aligner/command_line/align.py", line 186, in <module>
action='store_true')
File "aligner/command_line/align.py", line 151, in validate_args
def validate_args(args):
Exception: The language '/home/kranti/spanishmodel' is not currently included in the distribution, please align via training or specify one of the following language names: english.
Failed to execute script align
kranti@kranti:~/Montreal-Forced-Aligner-master$
POSSIBILITY :
YOU MAY HAVE UNZIPPED THE PRETRAINED MODEL.DO NOT UNZIP IT USE IT AS IT IS.
kranti@kranti:~/Montreal-Forced-Aligner-master$ bin/mfa_align -t ~/output -v ~/MFATestWav ~/SpanishDict/Spanishdict.txt ~/SpanishModel/spanish.zip spanish
Setting up corpus information...
SpanishModel CONTAINS spanish.zip DOWNLOADED FROM PRETRAINED MODEL WEBSITE.
ERROR NUMBER 4 :
DESCRIPTION :
PronunciationAcousticMismatchError: There were phones in the dictionary that
do not have acoustic models:
kranti@kranti:~/Montreal-Forced-Aligner-master$ bin/mfa_align -t ~/output -v ~/MFATestWav ~/SpanishDict/Spanishdict.txt ~/SpanishModel/spanish.zip spanish
Setting up corpus information...
Number of speakers in corpus: 1, average number of utterances per speaker: 1.0
Traceback (most recent call last):
File "aligner/command_line/align.py", line 186, in <module>
action='store_true')
File "aligner/command_line/align.py", line 146, in validate_args
pretrained_dir = os.path.join(root_dir, 'pretrained_models')
File "aligner/command_line/align.py", line 88, in align_corpus
shutil.rmtree(args.output_directory, ignore_errors=True)
File "aligner/models.py", line 129, in validate
raise (PronunciationAcousticMismatchError(missing_phones))
aligner.exceptions.PronunciationAcousticMismatchError: There were phones in the dictionary that do not have acoustic models:
Failed to execute script align
kranti@kranti:~/Montreal-Forced-Aligner-master$
SOMETHING LIKE ABOVE
POSSIBILITY :
YOU MAY NOT HAVE GLOBAL PHONE DICTIONARY IN SPANISH
MFA SPANISH G2P MODEL ARE TRAINED ON GLOBAL PHONE
DICTIONARY. MFA WON’T KNOW WHAT PHONE IN THE DICTIONARY
