| A00-1006 | input was off-line , i.e. , a | transcription | of dialogues , which was encoded |
| A00-1012 | ASR text with no errors in the | transcription | . The remaining three subjects |
| A00-2017 | conducted that use the phonetic | transcription | of the words to generate confusion |
| A00-1012 | identifying sentence boundaries in the | transcriptions | produced by automatic speech |
| A00-1044 | 130,000 words ) . Because the test | transcriptions | were created by humans , they |
| A00-1044 | used for this experiment was the | transcriptions | of the second 100 hours of the |
| A00-2028 | outputs a potentially errorful | transcription | of what it believes the caller |
| A00-1012 | not be relied upon as accurate | transcriptions | of the news program . The spoken |
| A00-2017 | first experiment ( Table 4 ) , the | transcription | of each word is given by the |
| A00-1044 | vocabulary speech recognizers doing | transcription | . The reason is that speech lexicons |
| A00-2038 | phonetic sequences presupposes | transcription | of sounds into discrete phonetic |
| A00-2029 | transcribed by hand and these | transcriptions | automatically compared to the |
| A00-1044 | without . We tested on the human | transcription | ( 0 % WER ) and the ASR ( 15 |
| A00-1012 | been removed . The texts were | transcriptions | of two editions of the news program |
| A00-2016 | presents examples in a romanized | transcription | . In sentence ( 1 ) for example |
| A00-1012 | more information than just the | transcription | . Under these conditions it is |
| A00-2040 | database which have a phonetic | transcription | . After several development cycles |
| A00-2028 | NLU ) module takes as input a | transcription | of the user 's utterance from |
| A00-1002 | technical terms by means of a direct | transcription | of productive endings and a slight |
| A00-1044 | noted that because the data are | transcriptions | of speech , no version of the |
