A new procedure to improve accuracy in an empirical transfer function measurement method is proposed for investigating speech sound propagation. In our previous work, vowel dependent behavior of empirical transfer functions from a lip reference point to observation points around a speaker's head was found. The accuracy of the method was also evaluated by using references obtained using a HATS and an M-sequence that revealed significant accuracy degradations in higher frequency range due to low speech energy. The proposed method solves this problem by introducing logarithmic temporal manipulation and low-pass filtering. The proposed method was tested using 186 vocalizations of sustained Japanese vowels. Test results indicated that the proposed method reduced standard deviations to 80% in gain estimation, 33.8% in weighted group delay estimation, and 20% in duration estimation, respectively, in frequency regions higher than 10 kHz. Detailed implementation aspects are also discussed.
|Journal||European Signal Processing Conference|
|Publication status||Published - 1 Dec 2006|
|Event||14th European Signal Processing Conference, EUSIPCO 2006 - Florence, Italy|
Duration: 4 Sep 2006 → 8 Sep 2006