| KRATKA VSEBINA | V članku opisujemo pristope, s katerimi lahko računalniško tvorimo zvoke, ki jih slišimo kot prostorske, torej prihajajoče iz določene smeri in z določene oddaljenosti. Pri takšnem zaznavanju v realnosti odigra pomembno vlogo t. i. impulzni odziv glave, kizdružuje vpliv uhljev in oblike glave. Računalniško tvorimo umetne zvoke tako, da te vplive že vsebujejo, zato jih je treba poslušati s slušalkami. Tako lahko simuliramo poslušanje zvočnega izvira na odprtem ali pa v zaprtem prostoru, ki ima določene dimenzije ter odbojne lastnosti sten in stropa. V obeh primerih obravnavamo situaciji, ko izvir in poslušalec mirujeta in ko se izvir giblje glede na mirujočega poslušalca. Pri tem smo uporabili standardizirane impulzne odzive človeške glave, ki so nam jih odstopili na univerzi Wisconsin-Madison v ZDA. Umetne 3D zvočne signale smopredvajali naključno izbranim poskusnim osebam. Ugotavljati so morale, iz katere smeri slišijo predvajane zvoke ali v kateri smeri se zvočilo giblje.Njihovi odgovori kažejo na prostorsko ločljivost ušesa med 5° in 10°. Poskusne osebe so tudi teže ločevale med smerema levo-spredaj in levo-zadaj. Prav tako jim je delalo težave gibanje zvočila pred njimi, torej spredaj, ki so ga zamenjevale z gibanjem zadaj. // When perceiving the sound, ears act as some kind of a filter that transforms the incoming signal with respect to the direction in which the signal source is located.A person can determine the direction of the sound just by listening to it. The sensation of the direction can be synthesized by transforming the signal in an appropriate way to obtain the so called 3D sounds. Computer synthesis of 3D sounds makes use of inter-aural delay of sound waves reaching one ear before the other. The brain detects this delay as a certain horizontal direction of arrival in space, measured by azimuth. However, there are also the elevation and distance in space, which place a sound source at a definite space point. For locating it thoroughly in 3D, the head related impulse response (HRIR) plays the most significant role. In Section 3, Eq. (3) defines the inter-aural delay, whereas Eqs. (2) and (4) explain how an audible signal can be convolved out of a mono sound source and HRIR. Section 3 reveals fundamentals and essentials for synthesis of 3D sound in open spaces. Subsection 3.1 applies to a still standing source. In the sequel, it is extended to moving sources. There, the main idea is based on cutting the source trajectory into successive intervals which are considered stationary. The sound at such intervals is generated according to the rules for still standing sources. It is also stressed that the convolution with HRIR must be computed with overlapping to avoid the transition effects (Eqs. (5) and (6). 3D computer sound generation in closed spaces is explained in Section 4. A model description based on geometry is introduced for the virtual room (Eq. (7)). Every wall reflects the sound similar to a mirror. The source image location is determined by Eqs. from (8) to (12). Additionally, the audibility of virtual sound sources must be verified, which is explained in Subsection 4.3 (Eqs. from (13) to (16)). Only the audible echoes are added to the generated 3D sound. Eq. (17) shows how to generate 3D sound heard in a closed space from a still standing source. Introduction of moving sources relates the moving sources as explained for open spaces. Section 5 reveals the results of practical experiments. The computer generated 3D sounds werelistened to by test persons. They were asked to describe their perception of the sound source location. The main conclusions confirm that human audible spatial resolution lies between 5° and 10°, the sense of direction is more accurate when one can compare two successive sounds at different locations, while several persons have troubles in mistaking the front locations with the rear ones. Finally, we would like to acknowledge the generous and kind help of Mrs. Sophia Chiron Stevens from the Waisman Center at the University Wisconsin-Madison, USA, who provided the head related impulse responses recorded in their laboratories. |