Robust automatic time alignment of orthographic transcriptions with unconstrained speech

摘要: A method for automatic time alignment of orthographically transcribed speech using supervised speaker-independent recognition based on the orthographic transcription, an online dictionary, and HMM phone models is presented. This successfully aligns transcriptions with in unconstrained 5 to 10 min conversations collected over long-distance telephone lines. It requires minimal manual processing generally produces correct alignments despite challenging nature data. The robustness efficiency make it a practical tool very large corpora. >