Abstract
Ensuring the best quality and performance of modern speech technologies, today, is possible based on the widespread use of machine learning methods. The idea of this project is to study and implement an end-to-end system of automatic speech recognition using machine learning methods, as well as to develop new mathematical models and algorithms for solving the problem of automatic speech recognition for agglutinative (Turkic) languages. Many research papers have shown that deep learning methods make it easier to train automatic speech recognition systems that use an end-to-end approach. This method can also train an automatic speech recognition system directly, that is, without manual work with raw signals. Despite the good recognition quality, this model has some drawbacks. These disadvantages are based on the need for a large amount of data for training. This is a serious problem for low-data languages, especially Turkic languages such as Kazakh and Azerbaijani. To solve this problem, various methods are needed to apply. Some methods are used for end-to-end speech recognition of languages belonging to the group of languages of the same family (agglutinative languages). Method for low-resource languages is transfer learning, and for large resources – multi-task learning. To increase efficiency and quickly solve the problem associated with a limited resource, transfer learning was used for the end-to-end model. The transfer learning method helped to fit a model trained on the Kazakh dataset to the Azerbaijani dataset. Thereby, two language corpora were trained simultaneously. Conducted experiments with two corpora show that transfer learning can reduce the symbol error rate, phoneme error rate (PER), by 14.23 % compared to baseline models (DNN+HMM, WaveNet, and CNC+LM). Therefore, the realized model with the transfer method can be used to recognize other low-resource languages.
Highlights
Automatic speech recognition systems began to develop dynamically with the rapid development of computing technologies
A speech corpus was assembled for the Kazakh language in the amount of 400 hours of speech, and for the Azerbaijani language, a speech corpus was formed amounting to 70 hours of speech
The joint connectionist temporal classification (CTC) attention models are trained based on the extracted features through the negative matrix factorization (NMF) algorithm
Summary
Automatic speech recognition systems began to develop dynamically with the rapid development of computing technologies. Accurate results have been achieved in the field of speech recognition, with many models and methods used in commercial applications, justifying their use in these directions. Among the commercial applications for speech recognition, first is the introduction of call centers or interactive voice response (IVR) systems for automatic access to information, speech chatbots, etc. Call centers have implemented intelligent voice assistants that generate user questions in natural language, and the response is synthesized by the system in the user’s language. Primary automatic speech recognition systems consist of three modules: decoding, acoustic, and language models. The modular subsystem for speech recognition mainly consists of independent modules, and even the acoustic model depends on the HMM model as well as GMM models, which, in many cases, correspond to the pronunciation unit [1]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
