<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">On 4/23/2013 7:20 PM, E wrote:<br>
</div>
<blockquote
cite="mid:8D00EA9618D72DD-71C-611D@webmail-d166.sysops.aol.com"
type="cite"><font size="2" color="black" face="arial">Hello all,
<div><br>
</div>
<div>I have n-best results <span style="font-size: 10pt; ">from
HTK </span><span style="font-size: 10pt; ">for lots of wav
files in a single MLF file. I want to compute nbest-error
rate using SRILM toolkit. I would like to know - </span></div>
<div><br>
1. Is there any direct way to convert HTK n-best MLF format
into SRILM n-best format (<a moz-do-not-send="true"
href="http://www.speech.sri.com/projects/srilm/manpages/nbest-format.5.html"
style="font-size: 10pt; ">http://www.speech.sri.com/projects/srilm/manpages/nbest-format.5.html</a><span
style="font-size: 10pt; ">). Or should I write a script for
that?</span></div>
</font></blockquote>
<font size="2"><font face="arial"><font size="2">S</font>orry, there
is no standard conversion tool <font size="2">I am aware of</font>.
I<font size="2">f you write one you <font size="2">should share
it with the list.</font></font></font></font><br>
<blockquote
cite="mid:8D00EA9618D72DD-71C-611D@webmail-d166.sysops.aol.com"
type="cite"><font size="2" color="black" face="arial">
<div><span style="font-size: 10pt; "><br>
</span></div>
<div><span style="font-size: 10pt; ">2. What does the n-best
error actually mean in SRILM? Suppose I have 10 hypotheses
for each wav file. I am familiar with usual way of reporting
errors using WER.</span></div>
</font></blockquote>
<blockquote
cite="mid:8D00EA9618D72DD-71C-611D@webmail-d166.sysops.aol.com"
type="cite"><font size="2" color="black" face="arial">
<div><span style="font-size: 10pt; "><br>
</span></div>
<div><span style="font-size: 10pt; ">Does nbest correctness mean
the number of words in reference that occur ANYWHERE in the
nbest-list? </span><span style="font-size: 10pt; ">Does
nbest deletions mean the number of words in reference that
occur NOWHERE in the nbest-list?</span><span
style="font-size: 10pt; "> What does nbest substitutions
mean?</span></div>
</font></blockquote>
<font size="2"><font face="arial">nbest<font size="2"> error is the
lowe<font size="2">st WER achievable by picking the <font
size="2">bes<font size="2">t</font> hypothesis (the one
giving the lowe<font size="2">st number of errors) </font>from
each nbest list<font size="2">. <font size="2">That's
why it<font size="2">'s also called the<font size="2">
<font size="2">"<font size="2">or<font size="2">acle"
error rate, as<font size="2"> an <font
size="2">oracle magical<font size="2">ly
told you which hypothes<font size="2">is
to pick to give the best result.<br>
<font size="2"><font size="2">The
number of deletions,
substitutions, etc. in this con<font
size="2">text is that <font
size="2">the <font size="2">number
of d<font size="2">eleted,
substituted, etc. ,
words <font size="2">relative
to the reference found
in that <font
size="2">or<font
size="2">acle
hypothesis.</font></font></font></font></font></font></font></font></font><br>
<br>
<font size="2">Andreas<br>
<br>
</font></font></font></font></font></font></font></font></font></font></font></font></font></font></font></font></font><br>
</body>
</html>