Requirements and Motivations of Low-Resource Speech Synthesis for Language Revitalization

Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 7346 - 7359 May 22-27, 2022 c⃝2022 Association for Computational Linguistics Requirements and Motivations of Low-Resource Speech Synthesis for Language Revitalization Aidan Pine1 aidan.pine@nrc.ca Dan Wells2 dan.wells@ed.ac.uk Nathan Thanyehténhas Brinklow3 nathan.brinklow@queensu.ca Patrick Littell1 patrick.littell@nrc.ca Korin Richmond2 korin.richmond@ed.ac.uk Abstract This paper describes the motivation and devel- opment of speech synthesis systems for the pur- poses of language revitalization. By building speech synthesis systems for three Indigenous languages spoken in Canada, Kanien’kéha, Gitksan & SENĆOŦEN, we re-evaluate the question of how much data is required to build low-resource speech synthesis systems featur- ing state-of-the-art neural models. For ex- ample, preliminary results with English data show that a FastSpeech2 model trained with 1 hour of training data can produce speech with comparable naturalness to a Tacotron2 model trained with 10 hours of data. Finally, we mo- tivate future research in evaluation and class- room integration in the field of speech synthe- sis for language revitalization. 1 Introduction There are approximately 70 Indigenous languages spoken in Canada, from 10 distinct language fam- ilies (Rice, 2008). As a consequence of the resi- dential school system and other policies of cultural suppression, the majority of these languages now have fewer than 500 fluent speakers remaining, most of them elderly. Despite this, interest from students and parents in Indigenous language edu- cation continues to grow (Statistics Canada, 2016); we have heard from teachers that they are over- whelmed with interest from potential students, and the growing trend towards online education means many students who have not previously had access to language classes now do. Supporting these growing cohorts of students comes with unique challenges for languages with few fluent first-language speakers. A particular concern of teachers is to provide their students with opportunities to hear the language outside of 1National Research Council Canada 2University of Edinburgh 3Queen’s University class. Text-to-speech synthesis technology (TTS) shows potential for supplementing text-based lan- guage learning tools with audio in the event that the domain is too large to be recorded directly, or as an interim solution pending recordings from first- language speakers. Development of TTS systems in this context faces several challenges. Most notable is the usual assumption that neural speech synthesis models re- quire at least tens of hours of audio recordings with corresponding text transcripts to be trained ade- quately. Such a data requirement is far beyond what is available for the languages we are con- cerned with, and is difficult to meet given the lim- ited time of the relatively small number of speak- ers of these languages. The limited availability of Indigenous language speakers also hinders the sub- jective evaluation methods often used in TTS stud- ies, where naturalness of synthetic speech samples is judged by speakers of the language in question. In this paper, we re-evaluate some of these chal- lenges for applying TTS in the low-resource con- text of language revitalization. We build TTS sys- tems for three …

Content truncated to 500 words