Constructing a speech audio–video corpus by aligning long segments of speech and text
Ivan Karpukhin, Anton Konushin · Moscow University Computational Mathematics and Cybernetics · 2017
A new algorithm for aligning text with speech audio signals having lengths of up to several hours is proposed. The algorithm allows its quality to be effectively evaluated. The requirements on the acoustic model are not very demanding. The algorithm can be used to design an audio–video course for learning the Russian language.