Training Effects on Speech Production Using a Hands-Free Electromyographically Controlled Electrolarynx

Abstract

Our findings suggest that some participants with laryngectomies and vocally normal individuals can learn to produce hands-free speech using the EMG-EL device within a few hours and that significant additional gains in device control (particularly pitch modulation) are attainable through subsequent training sessions.

Vocal frequency estimation and voicing state prediction with surface EMG pattern recognition

“…A further study defined a universal, linear positive transform function between EMG activity level and output pitch and explained it to normal participants. After moderate training, participants were able to regulate their infrahyoid muscle activity to consistently produce intonational contrasts that were large enough to distinguish between declarative and interrogative intentions [5]. However, this study focused on infrahyoid control in a broad sense, and participants…”

“…There is a notable absence of any consideration for variation in the frequency components of infrahyoid muscle EMG with fundamental frequency. Participants in the study conducted by Goldstein et al [5] were able to elevate infrahyoid muscle activity to increase pitch. This is in contrast to the role of the infrahyoid muscles in normal speech and is evidence of a new pitch regulation mechanism being assimilated.…”

“…Voice termination was therefore achieved passively. Since termination is normally an active process involving muscle contraction, EMG activation would be prolonged past the desired termination point, resulting in an undesirable increase in VTT (voice termination time) [5]. In addition, requiring users to learn additional skills augments the cognitive load needed to operate the device.…”

“…Saikachi et al [14] used the amplitude of electrolaryngeal speech to modulate the fundamental frequency. While these may be carried out in real time and results show improved speech quality, the information contained within the prosodic features of speech are not regained and can often be obfuscated [5,14].…”

“…The former inhibits intelligibility and limits the paralinguistic content of speech, while the latter precludes two-handed tasks during conversation [5]. Variation in fundamental frequency during vocal communication contains a large amount of information about speaker emotion and intention (e.g.…”