Substitution of Vocal Folds for Voice Generation by Means of Intra-Oral Pulse Generator

Human beings vocal folds act as a carrier pulse (or pitch-pulse, as said) for normal voice, with that, other parameters such as oral cavity structure or its elements positioning, shape this signal in what we know as speech. This characterizes normal voice as signals with a high concentration of energy at the lower bands of the frequency spectrum [11]. This research defines and implements a device capable of generating a carrier signal (pitchpulse) capable of, eventually, acting as a substitute for the vocal folds. This device generates a constant low frequency periodic pulse that allows the user to shape it when used intra-orally, just as one would do for normal speech creation. This research also evaluates the results of using such substituting device, and compares it with the results of normal speech. Keywords—non-audible speech; voice generation; spectral energy; carrier; frequency spectrum; vocal folds I. PURPOSE The purpose of this study is to design a device capable of generating a periodic signal, which would be able of transmitting non-audible speech. Such generated signal acts as a carrier, emulating real-voice when transmitting non-audible speech; this is, as vocal folds do. Non-audible speech refers to the changes in the oral cavity which modulate vocal tract impulses to generate the voice. In this case, the human vocal tract is substituted by an artificial pulse-train generator specifically designed for such task. II. PREVIOUS STUDIES IN THE FIELD Although studies in non-audible speech data have a wide history, efforts are mainly centered on transmission of this data, or even in recognizing this data for later voice generation [8] and [10]. Vocal folds substitute devices is still an unpopulated area but has some precedents in studies achieved by N. MacLeod [14] for telephony or Katz et al. [13] for a larynx substitution device. However, the purpose of this study is to substitute vocal folds as they are, and enable the subject to create voice without its vocal folds. More recently, there has been an approach by A. Passos and Prof. Y. Takeuchi on the use of extra-oral low frequency signals as a substitute for the vocal folds [6]. Or researches on unvoiced signals properties like [2], [9] and [12], where the conclusion that in order to achieve pseudo-voice regeneration of unvoiced signals (whispering voice), only the higher segment of the speech frequency spectrum is necessary has been reached. III. FURTHER STUDIES Taking into account that the purpose of this study is to design a carrier signal capable of carrying non-audible speech data, the scope of this research is still far from achieving the generation of voice from this non-audible data. However, thru enabling a subject to transmit such data, generation of voice can, eventually, be achieved by means of the methods described in this study. Such pseudo-voice generation can be achieved by, for example, playback of the autocorrelation of this signal. Pseudo-voice regeneration has, by far, much more implications and issues to be solved than this study intends to solve. IV. DEVICE FOR PULSE GENERATION A. Design The device features a low frequency pitch-pulse period which can be varied on demand to provide different types of pulse signals with differences in pulse duration, pulse frequency, etcetera. Figure 1. Schema of usage of the pulse-train generator as an intra-oral device The device has been designed for intra-oral use. The device has to be positioned at either the right or the left side of the oral cavity but always touching the cavity wall and facing to the inside of it. 978-1-4244-4131-0/09/$25.00 ©2009 IEEE