Executive Secretary
II Conferencia Internacional de Procesamiento de la Información
CIPI - IOTAI2019
Resumen
The present investigation will evaluate the impact of Mel Frequency Cepstral Coefficients (MFCC) and the Perceptual Linear Predictors (PLP) coefficients, in the word error rate (WER) of systems dedicated to Automatic Speech Recognition (ASR) for Spanish in scenarios with unknown noise levels using the state-of-the-art tool-kit Kaldi. Conclusions will be given on how the selection of the acoustic feature MFCC improves the WER for identical acoustic models, providing evidence on the robustness of the MFCC in the task of ASR in noisy scenarios.
Abstract
The present investigation will evaluate the impact of Mel Frequency Cepstral Coefficients (MFCC) and the Perceptual Linear Predictors (PLP) coefficients, in the word error rate (WER) of systems dedicated to Automatic Speech Recognition (ASR) for Spanish in scenarios with unknown noise levels using the state-of-the-art tool-kit Kaldi. Conclusions will be given on how the selection of the acoustic feature MFCC improves the WER for identical acoustic models, providing evidence on the robustness of the MFCC in the task of ASR in noisy scenarios.
Sobre el ponente
Ing. José Manuel Ramírez Sánchez
Graduated as Telecommunications and Electronics Engineer from the Universidad Tecnólogica de la Habana (UTH). He works in the Voice Group of the Advanced Technologies Application Center (CENATAV) dedicated to speech technologies, specifically in Automatic Speech Recognition (ASR) and Search on Speech (SOS). Member of the Cuban Association of Pattern Recognition.