Year: 2026
Type: conference-paper
Abstract: Maqolada nutqni avtomatik tanib olish (Automatic Speech Recognition, ASR) tizimlarida nutq signaliga dastlabki ishlov berishning asosiy bosqichlari — old kuchaytirish (pre-emfaza), signalni kadrlarga ajratish va oynalash, ovozli faollikni aniqlash (Voice Activity Detection, VAD), shovqindan tozalash hamda signal amplitudasini normallashtirish ko‘rib chiqilgan. Spektral ayirish, Viner filtri va veyvlet o‘zgartirishi asosidagi takomillashtirilgan shovqindan tozalash algoritmlari qiyosiy tahlil qilinib, ularning nutqni tanib olish aniqligiga ta’siri baholangan. Mel-chastotali kepstral koeffitsiyentlarni (MFCC) hisoblashdan oldingi dastlabki ishlov berish bosqichining ahamiyati asoslab berilgan.
Authors: Nurbek Nuritdinov, Elchinbek Nabiyev
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7164812425