Title: Applied Fletcher-Munson curve algorithm for improved voice recognition
Author: Bobby L. Tait
Address: University of Johannesburg, Cnr. of Beyers Naude Ave. and Kingsway Ave., Aucklandpark, 2006, South Africa
Abstract: The ability of a computing system, robot, or any man made system is to recognise humans commands and understand their instructions. Internationally, large research grants are awarded to this end. IBM Via Voice (http://www-1.ibm.com/software/pervasive/embedded_viavoice/), and efforts in the early versions Microsoft operating systems attempt to understand what a person said. The ultimate goal would be a PC that listens, interprets and understands human commands. Unfortunately, the effortless, exact voice recognition by (and almost colloquial conversation with) computers is still in the distant future. This paper considers a novel approach of improving the current voice recognition efforts of existing systems. The way that sound essentially operates is discussed; research by Fletcher-Munson (Vitz, 1972; Fletcher-Munson Curves, http://hyperphysics.phy-astr.gsu.edu/hbase/sound/eqloud.html) is integrated into a new algorithm used as a 'plug in', and implemented as a middle tier algorithm. The suggestions of this paper will serve as a step in the direction of allowing man made systems to interact with humans, using voice commands.
Keywords: voice recognition; biometrics; security; phon curve; sound characteristics; Fletcher-Munson curves; voice authentication; human-computer interaction; HCI; voice commands.
Int. J. of Electronic Security and Digital Forensics, 2012 Vol.4, No.2/3, pp.178 - 186
Submission date: 07 Dec 2011
Date of acceptance: 26 Apr 2012
Available online: 05 Aug 2012