Year: 2025
Type: article
Abstract: Automatic recognition of continuous speech is a key area focused on converting spoken words, phrases, and sentences into textual representations. This article addresses the challenges of recognizing continuous and mixed speech in the Uzbek language, presents neural network-based modeling solutions, and describes the phases of recognition using acoustic and language models. The processes of digitizing speech signals, extracting features, and implementing local and hypothetical recognition are discussed from an algorithmic perspective. The need for a specialized phonetic alphabet, morphological analyzer, transcription system, and acoustic/text corpora for developing an effective Uzbek ASR (Automatic Speech Recognition) system is emphasized. The proposed approach is universal and can be adapted to other languages as well. The research utilizes MFCCs, spectral coefficients, and machine learning methods based on neural networks. This work represents an important step toward developing Uzbek-language speech recognition systems.
Source: Neliti
Author: Nuritdinov, N. (N)
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7131993919