Tibetan speech corpus labeling method and system based on cooperative batch active learning

Abstract

The invention discloses a Tibetan speech corpus labeling method and a Tibetan speech corpus labeling system based on cooperative batch active learning. The system comprises a sample selection module,a manual labeling module, a labeling decision-making module, a labeling person evaluation module and a training set generation module. According to the method and the system, the construction of a sample evaluation function and the proving of the submodular function property of the sample evaluation function are solved through the adjacent optimal batch sample selection method, and the construction of a labeling decision function and the modeling of a labeling person evaluation model and a labeling person auxiliary learning model are solved through the labeling committee collaborative labelingmethod. In addition, the system disclosed by the invention can realize the functions inducing the optimal selection of a sample, the labeling and evaluation of a user, the sharing of labeling information and Tibetan speech knowledge, auxiliary learning of the labeling person and the like, so that the labeling quality of the Tibetan speech data is improved, and the construction of the speech corpus is accelerated.

Description

Technical field [0001] The invention relates to the field of speech recognition and corpus training, and in particular to a method and system for labeling Tibetan speech corpus based on collaborative batch active learning. Background technique [0002] In the field of speech recognition, traditional speech recognition algorithms (such as HMM, DBNs, ANN, DTW, etc.) use supervised learning methods to establish speech recognition models. In order to establish high-accuracy speech recognition models, this learning method requires a large number of bands. Annotating speech corpus, and annotating speech corpus is an extremely time-consuming and laborious task. Usually the time spent on labeling with words as the speech recognition unit is 10 times that of the actual audio sentence (for example, the labeling time for a one-minute speech sentence is close to 10 minutes), and the speech labeling with phoneme as the recognition unit It will reach 400 times the length of the speech sentenc...

AI Technical Summary

Problems solved by technology

Solve the construction of sample evaluation function and the proof of its submodularity property by approaching the optimal batch sample selection method

Through the collaborative labeling method of the labeling committee, the construction of the labeling decision function, the modeling of the labeler evaluation model and the labeler-assisted learning model are solved

Images