Air conduction + bone conduction = your own voice?
Gary C. Saleeby, George D. Allen, Robert B. Mahaffey, Thomas J. Wood · The Journal of the Acoustical Society of America · 1976
While subjects read a paragraph, a two-channel recording was made from an accelerometer strapped to the speaker's forehead (bone signal) and a microphone placed 5 cm from his ear (air signal). Each subject was then instructed to amplify or attenuate each channel on playback until the mixture most resembled his own voice. After adjustment, relative DBA measurements of bone vs air were taken from likewise prerecorded productions of isolated /n/ and /æ/. Intrasubject variability in these adjustments was low, indicating that the task could be performed. The mean-square intrasubject variability for /æ/ was, however, twice that for /n/, suggesting greater stability in the bone-air relationship of nasals. Intersubject variability was high and was three times greater for /æ/ than for /n/. The experimental system will be available for inspection and test at the session. Additional data will also be available regarding subjects' manipulations of the time relationship between the bone and air signals.