Presentation on theme: "Evaluation of Dummy-Head HRTFs in the Horizontal Plane based on the Peak-Valley Structure in One-degree Spatial Resolution Wersényi György SZÉCHENYI ISTVÁN."— Presentation transcript:
Evaluation of Dummy-Head HRTFs in the Horizontal Plane based on the Peak-Valley Structure in One-degree Spatial Resolution Wersényi György SZÉCHENYI ISTVÁN UNIVERSITY, Department of Telecommunications H-9026, Győr, Egyetem tér 1., Hungary Tel.: +3696613523 Fax: +3696429137 http://ta.sze.hu Abstract Dummy-heads are often used for standardized measurements where modeling of the human head and torso is relevant. Monaural Head-Related Transfer Functions of a Brüel & Kjaer dummy were measured in the horizontal plane in one-degree spatial resolution. Evaluation is made by plotting the peak-valley structure in frequency. Special frequency and spatial domains can be determined based on the variations of the HRTFs that are relevant to understand physical properties of the dummy-head in measurements and processes of human directional hearing. Symmetries and similarities of measured HRTFs help to scrutinize the perception of directional information in the monaural and binaural evaluation, the “noisy domain” in frequency and space where shadowing of the head occur as well as pinna-filtering. Introduction HRTF: Head-Related Transfer Function, transmission from free-field to the eardrum(s) Where P 1 is sound pressure at the eardrum and P 2 is the sound pressure in the middle of the head recorded with omnidirectional microphone Basic localization cue for directional hearing Measurement on real humans or on dummy-heads Reproduction through (equalized) headphones In listening tests are dummy-head HRTFs inferior to individual HRTFs Dummy-heads only partly model the human body and geometry Measurement HRTFs of a Brüel Kjaer 4128 dummy-head Anechoic room, white noise excitation Resolution: 1 degree horizontal, 5 degrees vertical Accuracy in the HRTFs: 1 dB Goal: –Investigating the peak-valley structure of the bare dummy-head’s HRTFs in the horizontal plane –Investigating the role of hair, cap, glasses, clothing etc.? The peak-valley structure in the horizontal plane Typical changes in the peak-valley structure as the sound source is moving in the horizontal plane Changes of about 1 dB will be evaluated Plotting of the nearby HRTFs in one degree resolution If the figure is „thin”: nearby HRTFs are very similar 0-30º Constant increase of the overall signal level up to 3-5 dB The 9 kHz peak increases with 7-9 dB 30-80º the signal level is constant The plotted HRTFs are very similar (though this is the most sensitive spatial region) Edges are very thin, only dips and peaks are changing 70-110º: 3 kHz peak and 4 kHz valley is falling down by 4 dB and 9 dB Thin edges and significant changes at peaks and valleys (5-7 dB) 90-140º: overall signal level decreases, especially the 4 kHz valley (20 dB) Local increase of signal level is in the „back direction” 200-320º: head-shadow area Minimum of the sensitivity: 250-260º (contralateral ear) Symmetry: ±20º to the median plane Summary Evaluation of measured dummy-head transfer functions (HRTFs) was made in the horizontal plane in one degree resolution. Spectral properties were investigated by plotting nearby HRTFs in the same figure. Different spatial regions can be determined based on these figures and based on properties such as thickness, deviations and variations of the peaks and valleys in the HRTFs. We have found the minimum of the monaural sensitivity of the hearing system on the contralateral side of the head, in the head-shadow area about 255 degrees. Furthermore, symmetries to the median plane both in frontal and in back directions are clearly visible in a ±20 degrees region. Due to the symmetry of the dummy-head, the same observation and conclusion can be drawn for the other ear as well and in real (binaural) listening situation these two regions overlap. If the ear is in the head-shadow area, only some low-frequency information can be evaluated. In general we can say that the edges of nearby HRTFs do not vary significantly. This indicates that changing of the azimuth does not really influence the peak-valley structure in the frequency (no shifting) only the height of the peaks and valleys. Fig.1. The head-related coordinate system Fig.2. Horizontal plane HRTFs from 0º and 30º. Fig.3. Ten horizontal plane HRTFs from 40º to 50º. Fig.4. Horizontal plane HRTFs from 70º and 110º. Fig.5. Horizontal plane HRTFs from 90º and 140º. Fig.7. Ten plotted horizontal plane HRTFs from 350º to 359º as well as from 0º to 9º for comaprison. Note the symmetry. Fig.6. Ten plotted horizontal plane HRTFs from the head-shadow area 250º to 259º (minimum of the sensitivity). Above 1600 Hz high frequency information is variable, below there is now significant difference.