Interspeech Shared Task: Automatic Speech Recognition for Non-Native Children’s Speech Forum

Go back to competition Back to thread list Post in this thread

> Potential problem with local/spellcorrect.awk - English data preparation

Hello,

One of the contestants has brought to our notice that local/spellcorrect.awk called by the data preparation script may fail on some systems. Error seen is:

awk: local/spellcorrect.awk: line 314: regular expression compile failed (missing operand)
?
fix_data_dir.sh: no utterances remained: not proceeding further

It fails when it is trying to remove question marks but as there are none in the .sup data files this is an unnecessary piece of code. If you have this problem remove lines:

lines 309-316
qn = match(loword,/\?/);
if (qn == length(loword)){
tmp = substr(loword,1,length(loword)-1);
loword = tmp;
} else if (qn > 0) {
gsub(/?/,"'",loword);
}

You can also email kate.knill@eng.cam.ac.uk for a copy of spellcorrect.awk without these lines.

Regards
Kate

Posted by: kateknill @ March 3, 2021, 10:57 a.m.
Post in this thread