Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This study evaluates the effectiveness of the Montreal Forced Aligner on Hindi-English code-mixed speech, addressing challenges such as phonemic boundary detection and speaker variation. By implementing bootstrapping strategies and training acoustic models on sentence-level code-mixed data, the authors achieved a mean alignment error of 4.15ms, significantly outperforming monolingual Hindi and isolated English models. The findings underscore the importance of tailored lexicon design and specialized training data for improving forced alignment in bilingual contexts.
Bootstrapping strategies can reduce forced alignment errors in Hindi-English code-mixed speech by a factor of ten compared to traditional models.
Code-mixed speech poses unique challenges to forced alignment: expanded inventories, orthographic errors, and speaker variation. We evaluate forced alignment of Hindi-English code-mixed speech using the Montreal Forced Aligner. We address 2 problems: (1) free variation involving native vs non-native pairs and (2) phonemic boundary detection for mid-utterance English words. Bootstrapping strategies substantially outperform unmodified lexicons. Acoustic models trained on sentence-level code-mixed data achieve a mean error of 4.15ms, ie. ten times lower than monolingual Hindi (38.18ms) or isolated English (37.58ms) alternatives. Principled lexicon design and code-mixed training data are both essential for reliable alignment of bilingual speech.