Speech sentence segmentation for indonesian automated speech recognition
The objective of this research was to create algorithm to do sentence segmentation on audio data without applying speech recognition process on the given audio data. Presently, sentence segmentation is done using language model posterior to speech recognition process, thus improving the readability of speech recognition result. Current thinking suggest that implementation of speech sentence segmentation preceding speech recognition process using signal features extracted from given audio data could improve the accuracy of speech recognition process.
B01180 | (wh) | Available |
No other version available