ISCA Archive Eurospeech 1991
ISCA Archive Eurospeech 1991

Up from trigrams! - the struggle for improved language models

Frederick Jelinek

The first experimental results in "large" vocabulary speech recognition were obtained in 1976 [ Bahl 78 ]. They involved continuous speech reading of. the so called Laser Patent Text. The test set was limited to sentences that were entirely composed of words belonging to a vocabulary of the 1000 most frequent words found in the training text.