knowledge sources integration framework for multilingual speech recognition for minority language

各种非物质文化表现形式脆弱易失，许多土著人民因语言和文化障碍而面临孤立。语音翻译对于土著社区保存其文化语言和克服语言障碍具有重要意义。本文主要研究语音翻译系统的一部分，即开发针对少数民族土著语言的语音识别系统。我们成功地收集了印度尼西亚现存的主要少数民族语言的数据，并分析了这些语言的特征。此外，通过将有效的知识库整合到印尼语的ASR基线中，成功地开发了少数民族语言的ASR。

Various intangible cultural expressions are fragile and easily lost, and many indigenous people face isolation due to language and cultural barriers. Speech translation is significant for indigenous communities to preserve their cultural language and overcome language barriers. This research focuses on one part of speech translation system which developing a speech recognition for minority indigenous languages. We were successfully collect the data from the main minority languages existing in Indonesia, and analyze the characteristics of those languages. Furthermore, by incorporating efficient knowledge source into Indonesian ASR baseline, the ASR of minority languages were successfully been developed.