Automatic speech recognition (ASR) systems are increasingly being developed for under-resourced languages, especially for use in multilingual spoken dialogue systems. We investigate different approaches to the acoustic modelling of Sepedi affricates for ASR. We determine that it is possible to model various of these complex consonants as a sequence of much simpler sounds. This approach reduces the Sepedi phoneme inventory from 45 to 32, resulting in simpler dictionary development and transcription processes, as well as more accurate acoustic modelling.
Reference:
Modipa, T, Davel, M and De Wet, F. Acoustic modelling of Sepedi affricates for ASR. 2010 Annual Research Conference of the South African Institute for Computer Scientists and Information Technologists (SAICSIT 2010), Bela Bela, South Africa, 11-13 October 2010
Modipa, T., Davel, M., & De Wet, F. (2010). Acoustic modelling of Sepedi affricates for ASR. ACM Digital Library. http://hdl.handle.net/10204/5631
Modipa, T, M Davel, and Febe De Wet. "Acoustic modelling of Sepedi affricates for ASR." (2010): http://hdl.handle.net/10204/5631
Modipa T, Davel M, De Wet F, Acoustic modelling of Sepedi affricates for ASR; ACM Digital Library; 2010. http://hdl.handle.net/10204/5631 .
2010 Annual Research Conference of the South African Institute for Computer Scientists and Information Technologists (SAICSIT 2010), Bela Bela, South Africa, 11-13 October 2010. This is an ABSTRACT ONLY
