Hi,
I'm new to kaldi and was trying out the mini_librispeech recipe. Since I do not have a grid engine installed, I changed
queue.pl to
run.pl in the cmd.sh script and I am running the run.sh script step by step (like ./run.sh --stage 2). It executed properly till stage 6 but I am getting the following error in stage 7:
local/download_and_untar.sh: data part dev-clean-2 was already successfully extracted, nothing to do.
local/download_and_untar.sh: data part train-clean-5 was already successfully extracted, nothing to do.
steps/get_prons.sh --cmd run.pl data/train_clean_5 data/lang_nosp exp/tri3b
steps/get_prons.sh: exp/tri3b/ali.1.gz exists, so starting from alignments.
run.pl: 5 / 5 failed, log is in exp/tri3b/log/nbest_to_prons.*.log
And this is the log mentioned in the error message:
#linear-to-nbest "ark:gunzip -c exp/tri3b/ali.5.gz|" "ark:
sym2int.pl --map-oov 3 -f 2- data/lang_nosp/words.txt <data/train_clean_5/split5/5/text |" "" "" ark:- | lattice-align-words data/lang_nosp/phones/
word_boundary.int exp/tri3b/final.mdl ark:- ark:- | nbest-to-prons exp/tri3b/final.mdl ark:- "|gzip -c >exp/tri3b/prons.5.gz"
bash: line 1: linear-to-nbest: command not found
bash: line 1: lattice-align-words: command not found
bash: line 1: nbest-to-prons: command not found# Accounting: time=0 threads=1
# Ended (code 127)
Can somebody tell me what seems to be the problem here?