search

amirharati / kaldi-alligner

scripts to align a given wave to its transcription using trained models by Kaldi
32 stars 7 forks source link

ERROR (arpa2fst[5.5.958~1-57f8d]:Input():kaldi-io.cc:756) Error opening input stream temp/lang/words.txt #3

Open Shiro-LK opened 3 years ago

Shiro-LK commented 3 years ago

Hello,

I am trying to launch the example in the readme.txt, unfortunately I don't have any files in the folder temp/lang which make me having this error :

FileNotFoundError: [Errno 2] No such file or directory: 'temp/lang/words.txt' temp/trans.txt: No such file or directory cat: temp/custom.lm: No such file or directory arpa2fst --disambig-symbol=#0 --read-symbol-table=temp/lang/words.txt - temp/lang/G.fst ERROR (arpa2fst[5.5.958~1-57f8d]:Input():kaldi-io.cc:756) Error opening input stream temp/lang/words.txt

[ Stack-Trace: ] /home/Shiro/kaldi/src/lib/libkaldi-base.so(kaldi::MessageLogger::LogMessage() const+0xb42) [0x7f20a5c98732] arpa2fst(kaldi::MessageLogger::LogAndThrow::operator=(kaldi::MessageLogger const&)+0x21) [0x55c95b2f3367] /home/Shiro/kaldi/src/lib/libkaldi-util.so(kaldi::Input::Input(std::__cxx11::basic_string<char, std::char_traits, std::allocator > const&, bool*)+0xbe) [0x7f20a5ebab08] arpa2fst(main+0x596) [0x55c95b2f1aa0] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe7) [0x7f20a4f8ebf7] arpa2fst(_start+0x2a) [0x55c95b2f142a]

kaldi::KaldiFatalErrormkgraph.sh: expected temp/lang/L.fst to exist Traceback (most recent call last): File "extract_likelihood_per_frame.py", line 26, in main() File "extract_likelihood_per_frame.py", line 21, in main fo.write(lpf + "\n") UnboundLocalError: local variable 'lpf' referenced before assignment lattice-1best ark:- ark:- lattice-align-words-lexicon temp/lang/phones/align_lexicon.int exp/tdnn_7b_chain_online/final.mdl 'ark:gunzip -c temp/out/lat.1.gz|' ark:- ERROR (lattice-align-words-lexicon[5.5.958~1-57f8d]:Input():kaldi-io.cc:756) Error opening input stream temp/lang/phones/align_lexicon.int

[ Stack-Trace: ] /home/Shiro/kaldi/src/lib/libkaldi-base.so(kaldi::MessageLogger::LogMessage() const+0xb42) [0x7ff1915ee732] lattice-align-words-lexicon(kaldi::MessageLogger::LogAndThrow::operator=(kaldi::MessageLogger const&)+0x21) [0x55fb185d7855] /home/Shiro/kaldi/src/lib/libkaldi-util.so(kaldi::Input::Input(std::__cxx11::basic_string<char, std::char_traits, std::allocator > const&, bool*)+0xbe) [0x7ff191ab8b08] lattice-align-words-lexicon(main+0x273) [0x55fb185d579d] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe7) [0x7ff1906c5bf7] lattice-align-words-lexicon(_start+0x2a) [0x55fb185d544a]

kaldi::KaldiFatalErrornbest-to-ctm ark:- temp/out/align.ctm LOG (lattice-1best[5.5.958~1-57f8d]:main():lattice-1best.cc:103) Done converting 0 to best path, 0 had errors. LOG (nbest-to-ctm[5.5.958~1-57f8d]:main():nbest-to-ctm.cc:119) Converted 0 linear lattices to ctm format; 0 had errors. gzip: temp/out/lat.1.gz: No such file or directory lattice-1best --acoustic-scale=1 ark:temp/out/lat.1 ark:temp/out/1best.lats LOG (lattice-1best[5.5.958~1-57f8d]:main():lattice-1best.cc:103) Done converting 0 to best path, 0 had errors. nbest-to-linear ark:temp/out/1best.lats ark,t:temp/out/1.ali LOG (nbest-to-linear[5.5.958~1-57f8d]:main():nbest-to-linear.cc:89) Done 0 n-best entries, 0 had errors. ali-to-phones --ctm-output exp/tdnn_7b_chain_online/final.mdl ark:temp/out/1.ali temp/out/phone_alined.ctm ERROR (ali-to-phones[5.5.958~1-57f8d]:Input():kaldi-io.cc:756) Error opening input stream exp/tdnn_7b_chain_online/final.mdl

[ Stack-Trace: ] /home/Shiro/kaldi/src/lib/libkaldi-base.so(kaldi::MessageLogger::LogMessage() const+0xb42) [0x7f803f618732] ali-to-phones(kaldi::MessageLogger::LogAndThrow::operator=(kaldi::MessageLogger const&)+0x21) [0x558d462e3953] /home/Shiro/kaldi/src/lib/libkaldi-util.so(kaldi::Input::Input(std::__cxx11::basic_string<char, std::char_traits, std::allocator > const&, bool*)+0xbe) [0x7f803fae2b08] ali-to-phones(main+0x450) [0x558d462e1c1a] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe7) [0x7f803ea82bf7] ali-to-phones(_start+0x2a) [0x558d462e16ea]

kaldi::KaldiFatalErrorTraceback (most recent call last): File "scripts/convert_ctm.py", line 48, in main() File "scripts/convert_ctm.py", line 32, in main lines = [line.strip() for line in open(args.input_words)] FileNotFoundError: [Errno 2] No such file or directory: 'temp/lang/words.txt' Traceback (most recent call last): File "scripts/convert_ctm.py", line 48, in main() File "scripts/convert_ctm.py", line 32, in main lines = [line.strip() for line in open(args.input_words)] FileNotFoundError: [Errno 2] No such file or directory: 'temp/lang/phones.txt' copy-int-vector ark:temp/out/1.ali ark,t:temp/out/transids.txt LOG (copy-int-vector[5.5.958~1-57f8d]:main():copy-int-vector.cc:83) Copied 0 vectors of int32. show-transitions temp/lang/phones.txt exp/tdnn_7b_chain_online/final.mdl ERROR: SymbolTable::ReadText: Can't open file temp/lang/phones.txt ERROR (show-transitions[5.5.958~1-57f8d]:main():show-transitions.cc:56) Could not read symbol table from file temp/lang/phones.txt

[ Stack-Trace: ] /home/Shiro/kaldi/src/lib/libkaldi-base.so(kaldi::MessageLogger::LogMessage() const+0xb42) [0x7fa88fff3732] show-transitions(kaldi::MessageLogger::LogAndThrow::operator=(kaldi::MessageLogger const&)+0x21) [0x55dc41c5b509] show-transitions(main+0x1c3) [0x55dc41c5ae0d] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe7) [0x7fa88f2e9bf7] show-transitions(_start+0x2a) [0x55dc41c5ab6a]

Am I the only one to have this issue ? Do I need to install or download something excepted Kaldi and SRILM?

amirharati commented 3 years ago

Have you download the models using: sh download_extract.sh also you should change your kaldi path (check the readme).

Also this code is relatively old and Kaldi has probably get updated in the meantime. I have not tested this code for a while and I don't have the time right now for testing with new kaldi, If you have the models and also updated the path and still get some error then perhaps you need to use older version of the Kaldi. I don't remember the version at time of writing this scripts, however you can approximate based on time of release.