Automatic Speech Recognition in Sanskrit: A New Speech Corpus and Modelling Insights

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 5039–5050 August 1–6, 2021. ©2021 Association for Computational Linguistics 5039 Automatic Speech Recognition in Sanskrit: A New Speech Corpus and Modelling Insights Devaraja Adiga1∗ pdadiga@iitb.ac.in Rishabh Kumar1∗ krrishabh@cse.iitb.ac.in Amrith Krishna2 ak2329@cam.ac.uk Preethi Jyothi1 pjyothi@cse.iitb.ac.in Ganesh Ramakrishnan1 ganesh@cse.iitb.ac.in Pawan Goyal3 pawang@cse.iitkgp.ac.in 1IIT Bombay, Mumbai, India; 2University of Cambridge, UK; 3IIT Kharagpur, WB, India Abstract Automatic speech recognition (ASR) in San- skrit is interesting, owing to the various lin- guistic peculiarities present in the language. The Sanskrit language is lexically productive, undergoes euphonic assimilation of phones at the word boundaries and exhibits variations in spelling conventions and in pronunciations. In this work, we propose the first large scale study of automatic speech recognition (ASR) in San- skrit, with an emphasis on the impact of unit selection in Sanskrit ASR. In this work, we release a 78 hour ASR dataset for Sanskrit, which faithfully captures several of the linguis- tic characteristics expressed by the language. We investigate the role of different acoustic model and language model units in ASR sys- tems for Sanskrit. We also propose a new mod- elling unit, inspired by the syllable level unit selection, that captures character sequences from one vowel in the word to the next vowel. We also highlight the importance of choos- ing graphemic representations for Sanskrit and show the impact of this choice on word er- ror rates (WER). Finally, we extend these in- sights from Sanskrit ASR for building ASR systems in two other Indic languages, Gujarati and Telugu. For both these languages, our ex- perimental results show that the use of pho- netic based graphemic representations in ASR results in performance improvements as com- pared to ASR systems that use native scripts.1 1 Introduction Sanskrit is a language with fairly advanced dis- ciplines of phonetics (Śiks ̣ā), prosody (Chandas), and grammar (Vyākaran ̣a). The language has a rich oral tradition and it tends to follow a phonemic-orthography resulting in systematic ∗Joint first author 1Dataset and code can be accessed from www.cse.iitb.ac.in/~asr and https://github. com/cyfer0618/Vaksanca.git. grapheme-phoneme correspondences. Connected speech leads to phonemic transformations in utter- acnes, and in Sanskrit this is faithfully preserved in writing as well (Krishna et al., 2018). This is called as Sandhi and is defined as the euphonic assimi- lation of sounds, i.e., modification and fusion of sounds, at or across the boundaries of grammatical units (Matthews, 2007, p. 353). Phonemic orthog- raphy is beneficial for a language, when it comes to designing automatic speech recognition Systems (ASR), specifically for unit selection at both the Acoustic Model (AM) and Language Model (LM) levels. Regardless of the aforementioned commonali- ties preserved in both the speech and text in San- skrit, designing a large scale ASR system raises several challenges. The Unicode encoding for the native scripts in Sanskrit, both in Roman and Devanāgari, does not preserve the correspondence with the phonemic encoding. Further, mapping the graphemes in Unicode to the corresponding phonemes either leads to ambiguity and redun- dancy or often …

Content truncated to 500 words