A bandit approach to curriculum generation for automatic speech recognition

Domain:

natural language processing

Record type:

paper

Creator:

KuzKumTye

Host:

The Automated Speech Recognition (ASR) task has been a challenging domain especially for low data scenarios with few audio examples. This is the main problem in training ASR systems on the data from low-resource or marginalized languages. In this paper we present an approach to mitigate the lack of training data by employing Automated Curriculum Learning in combination with an adversarial bandit approach inspired by Reinforcement learning. The goal of the approach is to optimize the training sequence of mini-batches ranked by the level of difficulty and compare the ASR performance metrics against the random training sequence and discrete curriculum. We test our approach on a truly low-resource language and show that the bandit framework has a good improvement over the baseline transfer-learning model.

Visit

arxiv.org

Tasks

automatic speech recognition speech processing

AUTOMATIC SPEECH RECOGNITION FOR AFAAN OROMO: A DEEP LEARNING APPROACH

Major Advisor: Jabesa Daba (Assist. Professor) Speech is one of the common styles of communication

A Yoruba Language Automatic Speech Recognition System using Deep Learning Approach

Reproducible Automatic Speech Recognition

The poster describes the architecture of one of the Sci-GaIA project "champion" use cases proposed f

Automatic speech recognition for under-resourced languages: A survey

(Impact-F 1.28 estim. in 2012) International audience no abstract

Developing a broadband automatic speech recognition system for Afrikaans

Automatic speech recognition system for court proceedings in Ethiopia Automatic speech recognition system for court proceedings in Ethiopia

This complete dataset supports the research article, "Automatic speech recognition system for court