Research on Tibetan Speech Recognition Based on the Am-do Dialect

Abstract

In China, Tibetan is usually divided into three major dialects: the Am-do, Khams and Lhasa dialects. The Am-do dialect evolved from ancient Tibetan and is a local variant of modern Tibetan. Although this dialect has its own specific historical and social conditions and development, there have been different degrees of communication with other ethnic groups, but all the abovementioned dialects developed from the same language: Tibetan. This paper uses the particularity of Tibetan suffixes in pronunciation and proposes a lexicon for the Am-do language, which optimizes the problems existing in previous research. Audio data of the Am-do dialect are expanded by data augmentation technology combining noise and reverberation, and the morphological characteristics and characteristics of the Tibetan language are further considered. According to the particularity of Tibetan grammar, grammatical features are used to optimize grammatical relationships and are combined with a language model, and the Am-do dialect is scored and rescored. Experimental results show that compared with the baseline, our proposed new lexicon and data augmentation technology yields a relative increase of approximately 3% in character error rates (CERs) and a relative increase of 3%-19% in the recognition rate of acoustic models and language models.

Improving Speech Enhancement Framework via Deep Learning

“…Therefore, after the speech input, the output sequence with the highest probability is selected, and after CTC decoding optimization, the final recognition result (x) is provided as output, where the operation formula is shown in Eq. (1).…”

“…In this study, deep learning was integrated into the development stage of the CTC-CNN model. The main problems are: (1) training usually must solve a highly nonlinear optimization problem, which may easily lead to many local minima during the training process of the network, and (2) a too-long training time may lead to overfitting results. In practical application, the system is stable, efficient, and general-purpose, and more than 97.5% of the recognition rate of noisy speech can be achieved.…”

“…Deep learning has been predominantly used in visual recognition, speech recognition, natural language processing, biomedicine, and other fields, where it has achieved excellent results. However, the performance of the acoustic model directly affects the accuracy and stability of the final speech recognition system, such that it is necessary to consider its establishment, optimization, and efficiency in detail [1]. The experiments in this study employed CTC-CNN, which exhibits a better performance than the earlier commonly employed Gaussian Mixture Model-Hidden Markov Model (GMM-HMM) acoustic model, to train the acoustic model.…”

Speech Recognition via CTC-CNN Model

“…Deep learning is predominantly used in visual recognition, speech recognition, natural language processing, biomedicine, and other fields, where it has achieved excellent results. However, in the field of speech recognition, the performance of the acoustic model directly affects the accuracy and stability of the final speech recognition system, such that it is necessary to consider its establishment, optimization, and efficiency in detail [1].…”