Automatic Speech Recognition and Access: 20 Years, 20 Months, or Tomorrow?
By Davis, Cheryl D.; Hearing Loss Magazine, Vol. 22, No. 4, pp. 11-14Publication Date: July/August 2001
Article highlights some breakthroughs in speech recognition technology, such as greater memory, which allows for larger dictionaries and faster processors. These features have greatly increased word-per-minute input and have decreased error rate. There are also some limitations of current speech recognition technology, as the article explains what is required for speech recognition technology to be more useful. For example, text would be more readable with commands such as "period," or "new paragraph" being recognized as punctuation rather than part of the text. Another limitation is that most speech recognition programs cannot distinguish between regional colloquialisms, as no two people say the same word the same way. Also, phrases are often spoken in such a way that syllables are spelled out in the wrong place - for example, "Coke is it" would be spelled "co ki sit." There is another problem with words sounding the same and having different meaning, such as pair, pear, and pare. The author foresees these issues to be remedied in the near future.
Published by: Hearing Loss Association of America (formerly Self Help for Hard of Hearing People) (Website:http://www.hearingloss.org)
