Some common mistakes Autor vlákna: Heinrich Pesch
| Heinrich Pesch Finsko Local time: 22:07 Člen (2003) finština -> němčina + ... | One of my favourites... | Jul 23, 2010 |
... is than and then! | | | Brian Young Spojené státy americké Local time: 12:07 dánština -> angličtina not the same | Jul 23, 2010 |
Than and then are not pronounced the same, although in everyday speech it may sound that way. I always try to make a clear distinction between the two. There and their, however, do sound exactly the same. | | | Nicole Schnell Spojené státy americké Local time: 12:07 angličtina -> němčina + ... In memoriam Speech recognition | Jul 23, 2010 |
Interesting point. I have never used it and I am curious how speech recognition deals with: hour (pronunciation: au(-ə)r) and our (pronunciation: au(-ə)r) for example. | |
|
|
Judith Kiraly Spojené státy americké Local time: 12:07 angličtina -> maďarština
The speech recognition software can take into account the context (3 words) based on a statistical model. | | |
Prayer and prayer are confusing to some. A local choir performing the well-known "Hear my prayer" pronounced 'prayer' in the context of a person praying, which was incorrect. I carefully pointed this out to the choir leader but to no avail - the same mistake was made in a performance a year later. Some people can english good. | | | Brian Young Spojené státy americké Local time: 12:07 dánština -> angličtina Judith is right | Jul 23, 2010 |
I have been using speech recognition whenever possible, and although I never thought about this as a potential problem, the program always seems to get it right (almost). Next time I will try a little test. If we use one hour on our project, then it will be less than their project over there. | | | Nicole Schnell Spojené státy americké Local time: 12:07 angličtina -> němčina + ... In memoriam Thanks, Judith! | Jul 23, 2010 |
Judith Kiraly wrote: The speech recognition software can take into account the context (3 words) based on a statistical model. Thanks for the info! | |
|
|
opolt Německo Local time: 21:07 angličtina -> němčina + ... This reminds of the phonetics course ... | Jul 23, 2010 |
... during my English studies. One of the sentences we had to practise aloud went like this: "If I had a hat." "If I had a hat." "If I had a hat." (The background is that in German, final consonants often turn into their hard, aspired versions if in final position, such that "Abend" is pronounced like "Abent", "Tag" as "Tak" , "ab" as "ap", etc. What we call "Auslautverhärtung". -- So this was one of those exercises intended to "de-learn" the habit.) We al... See more ... during my English studies. One of the sentences we had to practise aloud went like this: "If I had a hat." "If I had a hat." "If I had a hat." (The background is that in German, final consonants often turn into their hard, aspired versions if in final position, such that "Abend" is pronounced like "Abent", "Tag" as "Tak" , "ab" as "ap", etc. What we call "Auslautverhärtung". -- So this was one of those exercises intended to "de-learn" the habit.) We also had exercises to dinstiguish between "than" and "then" (mentioned earlier in the thread). -- (Oops, this is another one: "thread vs. threat".) I keep telling people that contrary to the almost universal belief in this country, English is a rather difficult language for German natives to learn, partly owing to its many phonetic subtleties and irregularities. But of course, no-one ever wants to believe me ▲ Collapse | | | Similar problem in OCR | Jul 25, 2010 |
That's not exclusive to voice recognition. Apparently it happens every time machines are expected to "understand" human language. Though there are several letter combinations that - depending on the font used - cause two letters to be merged, or one letter to be split, the most frustrating OCR flaw is to merge lowercase RNinto M, or to split a lowercase M into RN. A spellchecker will accept both, e.g. "dam" (DAM) and "darn" (DARN). Other common such things are: D ... See more That's not exclusive to voice recognition. Apparently it happens every time machines are expected to "understand" human language. Though there are several letter combinations that - depending on the font used - cause two letters to be merged, or one letter to be split, the most frustrating OCR flaw is to merge lowercase RNinto M, or to split a lowercase M into RN. A spellchecker will accept both, e.g. "dam" (DAM) and "darn" (DARN). Other common such things are: D into I) O (letter "o") into 0 (zero) I into | l (lowercase L) and 1 (one) Of course, good OCR software has its own internal "spellchecker" to fix most - but not all - of these. Things get worse when diacritics are involved. "ó" (letter O with an acute accent) is often a dead ringer for the number 6. So if accuracy is a top priority after a human-to-machine comversion, careful human reading of the text in some monospace font may be unavoidable.
[Edited at 2010-07-25 12:16 GMT] ▲ Collapse | | | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Some common mistakes Trados Studio 2022 Freelance | The leading translation software used by over 270,000 translators.
Designed with your feedback in mind, Trados Studio 2022 delivers an unrivalled, powerful desktop
and cloud solution, empowering you to work in the most efficient and cost-effective way.
More info » |
| CafeTran Espresso | You've never met a CAT tool this clever!
Translate faster & easier, using a sophisticated CAT tool built by a translator / developer.
Accept jobs from clients who use Trados, MemoQ, Wordfast & major CAT tools.
Download and start using CafeTran Espresso -- for free
Buy now! » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |