Talk:Speech analytics/Archive 1

Word Error Rate
"The best LVCSR engines may reach about 50% WER (word error rate). This is considered top of the line performance in today's standards.[weasel words] Notice however that this means that every second word you read in the output is wrong! Still, it may provide more then enough accuracy for statistical analytics. The quality of the phonetic engines is considerably lower. While hard to compare apples to apples, it is in the range of what would be 20% WER."

WER, like any error rate, is like a golf score -- lower is better, not worse. A 20% WER for phonetic engines would mean (roughly) 4 out of 5 words is correct. I'm not changing the text right now since there's no citation for either statistic anyway, but if they are correct, the section should be rewritten to state that phonetic engines perform better than LVCSR engines.12.38.201.65 (talk) 19:40, 10 February 2009 (UTC)

Citing WER for phonetic solutions is completely misleading. Phonetic solutions, by definition, do not produce words so Word Error Rates simply cannot be stated.

Not specific to telephone speech
The article currently defines speech analytics only in relation to telephone speech, and specifically contact centers. It needs rewording to clarify speech analytics is a general technique, applicable to any speech, aimed at finding out one or more of a number of bits of information about the speech, e.g. what was said, who was speaking, what languages were used, what emotional states the speakers were in etc etc.

The contact center applications can be used as examples, but shouldn't dominate the general definitions in the article. Howard Wright 11:45, 13 February 2007 (UTC) Howard Wright

List of vendors
Is a list of prominent vendors a useful addition to the article? How is prominence determined? How can such an addition be reliably sourced? Is such a list just spam? JackSchmidt (talk) 21:29, 24 July 2008 (UTC)

It is not a SPAM. Information is not SPAM. As long as the information is not biased, and tries to be complete and fair, it has value. Just becasue some bully monitoring this thread decided he doen't like it, doesn't make him right. Wikipedia is full of subjects that have a list of vendors. check out Spreadsheets. —Preceding unsigned comment added by Csmba (talk • contribs) 2008-08-01T21:52:46Z


 * The important thing is whether the information meets wikipedia's policies for inclusion. For instance, how do we decide which companies are listed?  Is there a magazine or industry newspaper that would have a list we could refer to? JackSchmidt (talk) 23:06, 1 August 2008 (UTC)


 * You would do well to read the relevant Wikipedia policies rather than to coddle the spammer. Even if a "reliable source" could be found it would certainly not be independent of the subject.  Furthermore, Wikipedia is not directory of vendors.  We should not be encouraging spammers such as User:Csmba from using Wikipedia as a platform to advertise for speech analytics vendors.
 * Finally, this is not an article. It is an advertisement masquerading as an article and should have been speedily deleted upon creation.  It was created by a WP:SPA (see Special:Contributions/Magnolia_Grandiflora) and the vast majority of edits have been to add spamlinks from anonymous accounts, particularly by Csmba anonymously (see Special:Contributions/209.77.216.34).  Please review the following spam edits before you revert the spam.
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=44376605
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=79106545
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=90748066
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=119694002
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=168484808
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=169743092
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=178382002
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=215190404
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=225919147
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=227537238
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=227705301
 * http://en.wikipedia.org/w/index.php?title=Speech_analytics&diff=next&oldid=228955272
 * -- DanielPenfield (talk) 20:49, 3 August 2008 (UTC)

By Csmba: Thanks for calling me a spammer. Now to the issue: 1. A spammer has something to gain. What am I gaining by publishing a list of vendors? commissions? 2. I am no lawyer and have no time to research the long rules and regulations. I just have time to use common sense. So I did a quick search and found that Spreadsheets has list of vendors. I can come up with many more examples. So can you explain why this is SPAM and it is ok in those postings? —Preceding unsigned comment added by Csmba (talk • contribs) 18:40, 7 August 2008 (UTC)

As for calling what I post (list of vendors) spam just becasue you don't know how to do fact validations, ( and I am sorry, I do not want to flame, I am just insulted), I don't understand that either. I thought this should be a source of information, a complilation, not a reference to other places on the web. some of this information comes from our head maybe? our own research? here, I found a reference, but you can say they are also not objective (who the hell is?) http://www.dmgconsult.com/services/speech/toc.asp find list of vendors they cover at point 23.

In any case, why are you the "good" police and me the "bad" spammer? why do I need to prove anything to you? don't you think that just as it is bad that one person will cause harm by posting spam or promotional, biased information, it is bad that one cop terrorize the information with his veto power? over using veto power is just as bad as posting biased information. it is also spam. —Preceding unsigned comment added by Csmba (talk • contribs) 19:13, 7 August 2008 (UTC) In any case, it makes no sense to me that one of the reasons why you call the list I posted of vendors "SPAM" is because I had no reference. How on earth would a reference make the list of vendors more or less spam? becasue YOU can police my post? as I said before, you are free to do your own research and decide that the list if not complete, and IMPROVE it. IMPROVE by adding content, not removing it. and don't be so condescending and rude to people you don't know and don't know their real motive. —Preceding unsigned comment added by Csmba (talk • contribs) 19:21, 7 August 2008 (UTC)

Why not discuss things?
I keep trying to improve this post, but it keeps being trimmed down. Guys, I am not trying to SPAM, please LMK how we can open a discussion so we don't go back and FW with this.... I want to make this topic better! I think specific names of providers is valid based on other wikipedia titles plus my personal opinion. lets discuss. —Preceding unsigned comment added by Csmba (talk • contribs) 2008-08-07T18:59:50Z