Toward a better understanding of the perception of self-produced speech

The subject of how humans hear their own voices while speaking has received sporadic attention. Previous investigations suggest that the voice is heard substantially via bone conduction, implying a different transfer function to the ear--perhaps a low-pass filter--than for purely air-conducted vocalizations. Determining the characteristics of the transfer function of self-produced speech might contribute to a better understanding of self-perception. Such information may have important implications for treating articulation disorders. The efficacy of training discrimination during the remediation of articulation disorders remains unclear, perhaps because clients are asked to discriminate either speech of others or recorded self-speech samples not readily recognized as their own. This study was conducted to determine preference for filtered self-monitored speech in a delayed feedback paradigm. Participants made immediate comparisons of their real-time speech to delayed-recorded filtered self-speech samples. Participants demonstrated a clear preference for low-pass filtered speech, supporting earlier findings, but we were unable to determine the transfer function itself.

As a result of this activity, the learner will be able to describe some reasons why an individual's tape recorded speech typically sounds different to him/her from his/her own, live speech. As a result of this activity, the learner will be able to describe some potential applications for an accurately-characterized transfer function for self-produced speech.