Noise Reduction Scheme for Speech Recognition in Mobile Devices
This paper proposes an efficient noise reduction scheme for speech recognition in mobile devices. Due to the limited capacity of mobile devices, the speech recognition system requires a noise reduction module processed with low computational intensity. For noise reduction in mobile devices, the proposed approach directly utilizes packet data estimated by a speech coder. In particular, we apply pitch information for comb filtering, a well-known noise reduction method.
KeywordsNoise reduction Speech recognition Mobile device Comb filtering
This research was supported by the Converging Research Center Program funded by the Ministry of Education, Science and Technology (2012K001341).
- 1.Lim MJ, Lee ES, Kwon YM (2012) Implementing mobile interface based voice recognition system. Computer Applications for Database, Education, and Ubiquitous Computing. 352:150–157Google Scholar
- 2.Kim HK (2000) Bitstream-based feature extraction for wireless speech recognition. In: International conference on ASSP, pp 1607–1610Google Scholar
- 3.Frazier RH (1975) An adaptive filtering approach toward speech enhancement. S.M.thesis, M.I.T., CambridgeGoogle Scholar
- 4.Frazier RH, Samsam S (1976) Enhancement of speech by adaptive filtering. In: International conference on ASSP, pp 251–253Google Scholar
- 5.Gardner W (1993) QCELP: a variable rate speech coder for CDMA digital cellular. Speech and audio coding for wireless and network applications. Kluwer Academic Pub., Dordrecht pp 77–84Google Scholar
- 6.Atal BS, Rabiner LR (1976) A pattern recognition approach to voiced-unvoiced-silence classification with applications to speech recognition. IEEE Trans Acoust Speech Signal Process 24:201–212Google Scholar
- 7.Hirsch HG, Pearce D (2000) The AURORA experimental framework for the performance evaluations of speech recognition systems under noisy conditions. In: ISCA Tutorial and Research WorkshopGoogle Scholar