Can not training with another language

[phamkhactu]

Thanks for your excellent working!

I want to training my g2p with other language, in my case is vietnamese

phạc ph a_T5 c2
num n u_T0 m2
rim r i_T0 m2
giẫn gi a3_T4 n2
toăm t oa2_T0 m2
lịu l iu_T5
cựi c u2i_T5
õng o_T4 ng2

I get error:

INFO:phonetisaurus-train:2023-09-06 17:51:21:  Checking command configuration...
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  Directory does not exist.  Trying to create.
INFO:phonetisaurus-train:2023-09-06 17:51:21:  Checking lexicon for reserved characters: '}', '|', '_'...
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  arpa_path:  train/model.o8.arpa
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  corpus_path:  train/model.corpus
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  dir_prefix:  train
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  grow:  False
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  lexicon_file:  /tmp/tmp53qaxdn7.txt
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  logger:  <Logger phonetisaurus-train (DEBUG)>
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  makeJointNgramCommand:  <bound method G2PModelTrainer._mitlm of <__main__.G2PModelTrainer object at 0x7fd3aae0ec40>>
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  model_path:  train/model.fst
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  model_prefix:  model
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  ngram_order:  8
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  seq1_del:  False
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  seq1_max:  2
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  seq2_del:  True
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  seq2_max:  2
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  verbose:  True
DEBUG:phonetisaurus-train:2023-09-06 17:51:21:  phonetisaurus-align --input=/tmp/tmp53qaxdn7.txt --ofile=train/model.corpus --seq1_del=false --seq2_del=true --seq1_max=2 --seq2_max=2 --grow=false
INFO:phonetisaurus-train:2023-09-06 17:51:21:  Aligning lexicon...
GitRevision: package
Loading input file: /tmp/tmp53qaxdn7.txt
Please provide a valid input file.
ERROR:phonetisaurus-train:2023-09-06 17:51:21:  Alignment failed.  Exiting.
Traceback (most recent call last):
  File "/home/tupk/anaconda3/envs/nlp/bin/phonetisaurus", line 8, in <module>
    sys.exit(main())
  File "/home/tupk/anaconda3/envs/nlp/lib/python3.8/site-packages/phonetisaurus/__main__.py", line 74, in main
    do_train(args, casing, env)
  File "/home/tupk/anaconda3/envs/nlp/lib/python3.8/site-packages/phonetisaurus/__main__.py", line 209, in do_train
    train(lexicon=lexicon, model_path=args.model, corpus_path=args.corpus, env=env)
  File "/home/tupk/anaconda3/envs/nlp/lib/python3.8/site-packages/phonetisaurus/__init__.py", line 121, in train
    subprocess.check_call(train_cmd, cwd=temp_dir_str, env=env)
  File "/home/tupk/anaconda3/envs/nlp/lib/python3.8/subprocess.py", line 364, in check_call
    raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['phonetisaurus-train', '--lexicon', '/tmp/tmp53qaxdn7.txt', '--seq2_del', '--verbose']' returned non-zero exit status 1.

But if I train with English lexicon, no problem

How can I fix it?
Thank you

[Pallas13]

Hello, some update this? I go develop a TTS Engine based in Flite for Android for my worked of High School, trained of g2p in Flite is much slower. I found this g2p but not know have bugs or have how trained for other language. Your lexicon is in UTF8?