ISCA Archive Eurospeech 1999
ISCA Archive Eurospeech 1999

Liftered forward masking procedure for robust digits recognition

Kaisheng Yao, Bertram Shi, Pascale Fung, Zhigang Cao

Using TI digits recognition experiments, we show that a combination of two dynamic speech features, Liftered Forward Masked (LFM) MFCC and 2-D cepstrum, can improve system robustness to additive Volvo noise while maintaining system per-formance comparable to standard MFCC features in clean conditions. Through experiments, we show that the information extracted by forward masking and by the 2D cepstrum are in some sense orthogonal. By combining the LFM MFCC and the 2-D cepstrum plus > 2-D cepstrum, we achieve a recognition rate above 90% on the TI connected digits task, even in additive Volvo noise condition with SNR as low as 0dB. This corresponds to a SNR gain over 30dB compared with standard MFCC plus dynamic and acceleration coefficients.


doi: 10.21437/Eurospeech.1999-636

Cite as: Yao, K., Shi, B., Fung, P., Cao, Z. (1999) Liftered forward masking procedure for robust digits recognition. Proc. 6th European Conference on Speech Communication and Technology (Eurospeech 1999), 2873-2876, doi: 10.21437/Eurospeech.1999-636

@inproceedings{yao99_eurospeech,
  author={Kaisheng Yao and Bertram Shi and Pascale Fung and Zhigang Cao},
  title={{Liftered forward masking procedure for robust digits recognition}},
  year=1999,
  booktitle={Proc. 6th European Conference on Speech Communication and Technology (Eurospeech 1999)},
  pages={2873--2876},
  doi={10.21437/Eurospeech.1999-636}
}