In our every day life we can perceive sounds coming from different directions. Our ear is able to define the direction on a sound and the distance to its source more or less precisely. So it is very desirable that sound reproducing systems would provide 3-dimension sound. Imagine that you watch on TV horrors movie with the main hero making his way through a graveyard in the dark full of mysterious sounds. But you are not so very afraid of those sounds because they do not surround you - the TV set is the source of them. As a result the effect of the movie is not so impressive as it might have been had it used 3D sound. On the other hand, using of 3D sound in the recording of musical pieces is not so obvious. Scrupulous recording of the 3D noise of the audience on a concert might have an unpredicted result. Some day you would turn around to stop blowing one's nose while listening to a 3D-sound demo record downloaded from Internet. However it is something other than a professional 3d sound recording.

3D sound is essential for computer games. Users and games developers have voted for 3D-sound technologies unanimously. It has found its reflection in Microsoft DirectX software that enables the creating of a 3D-sound with the help of Monster Sound 3D, Turtle Beach Montego A3DXStream or SoundBlaster Live sound cards. Generally saying, using of DirectX/DirectSound linked sound cards should provide arranging of virtual sound sources in any position. At the same time you need only two real sources (speakers or stereo phones) as a minimal but functional configuration. Thus you can improve a usual stereo sound from your computer speakers putting them virtually on the optimum distance from the hearer. Note that there is no need to move the speakers. The 3D sound processor will do it for you. 

Actually a stereo sound was the first step towards a multidimensional sound. But a usual stereo system or headphones could not grant a full satisfaction to a real audiophile. A stereo sound had an essential defect. It was flat and restricted to an angle between the directions to the speakers. So it did not sound that natural.

Quadrophony was the next step in the sound recording and reproducing. JVC CD-4, CBS SQ and Sansui QS non-compatible quadrophony systems were designed. But experts did not praise them too much. Most users noticed that they did not fully imitate a real sound. First of all quadrophony did not provide a round stereo panorama. The hearer perceived a usual stereo panorama in front and behind him. Secondly, all sound sources were lined up on the surface between the speakers. So there was no the third dimension and 3D sound altogether. After quadrophony failed the developers decided to increase the number of sound reproducing channels. As a result very multichannel "monsters" were born.

Dolby has become the leader in the development of mutichannel sound reproducing systems. Multichannel Surround Sound hardware by Dolby was installed in more than 14000 cinemas (according to http://www.dolby.com). When the time of a great demand for super multichannel devices passed Dolby invented a more ascetic system called Dolby Digital Surround. It consisted of only six speakers (left one, central, right, subwoofer, left and right surround - picture 3) and was meant for home use. 

It is ridiculous but Dolby Digital had amazing resemblance to a Russian ABC system designed in the Seventies. The main feature of both systems was the position of the rear speakers: they were not behind the hearer but strictly to the left and right of him. Such arrangement of speakers provided a round sound panorama. Naturally, there was no the third dimension too.

Binaural sound recording is the most faultless methodology of a real 3D sound imitation. In this case audio information is recorded by the microphones placed into the ears of either a human being or an artificial head simulating a human sense of hearing. That is the idea of the method.

The signals from each microphone are amplified by separate low frequency amps and played back by stereo headphones. Such system enables creating of the illusion of a live, natural sound. It brings the hearer to the room the sound coming from. But you need high quality stereo headphones and an artificial head made after your own head to appreciate its advantages in the full measure. It is recommended to use special microphones placed into your own ears to achieve a maximum effect (picture 4).

Note that only you yourself will appreciate the record made by the microphones placed into your ears. Other people can feel distinction between the sound imitation and its original because the characteristics of your (or other people's) sense of hearing might differ from the average.

Playing back of a binaural record via speakers can reduce to nothing all its advantages. Signals from the right and left speakers can get respectively to the left and right ears. As a result the sound distorts. It is possible to eliminate such a defect using special signal processing devices known as biphonic processors. They enable to obtain a binaural effect while listening to a binaural record via speakers. The record made by microphones is played back after being processed by biphonic processors where the value of the phased, delayed and equalized signals from the left and right channels is subtracted respectively from that of the right and left channels. Then the signals coming from the speakers to the ears are summed up so that the left and right ears can hear the signals from the left and right channels correspondingly. Therefore, a biphonic effect is similar to a binaural one. The way of the reproducing of a binaural record is the only difference. Although the zone where you can catch a biphonic effect is limited but being there you are able to know the distance to the sources of the sound and their position at the time of recording. A stereo sound gives you only an idea of the sound location on the line in between the speakers. Moreover, biphonic processors can extend the stereo base of a usual stereo record. Of course, a biphonic processor can be realized as a software based on the real time digital sound processing methods.

Real sound sources can be considered to be the points. So, they are mono sources. A stereo effect appears when the sound is perceived by the right and left ears simultaneously. Frequency characteristics of human ears depend on the azimuth-latitude location of the sound source and the distance to it. Besides they can change in a different way for each ear. The average frequency characteristics of the right and left ears are equal if only a sound source is in the surface of the symmetry of the head. On picture 6 one can see the difference of impulse and amplitude-frequency responses got from the microphones inside the ears when a sound source is off the surface of the symmetry.

Numerous data show the considerable alteration of frequency responses of each ear as long as the distance and direction to the sound source change. Different values of signal delays and amplitudes are most obvious. Apparently, a human brain analyzes the data to locate the sound source in 3D space. So, the feeling of the space location of the sound is linked with characteristics of the human sense of hearing (amplitude-frequency response and signal delay). If you know the average value of the characteristics of a human ear you are able to design digital filters to imitate its features. The more detailed is the space-frequency model of an ear, the more complicated are the filters. The imitation of an artificial 3D sound is the next step - signals from mono sources are processed by couples of digital filters (for the right and left ears) with parameters corresponding to the desired locations of the sound sources.

Affordable and very powerful digital signal processors (DSP) have appeared in the market for the last few years. AD21160 SHARC by Analog Devices (http://www.analog.com) is one of the most interesting of that kind. A 32-bit device with a 1000 MIPS calculation power can process integer and float point numbers. Its instructions enable software developers to fulfil both system and specific calculation tasks. C-compiler is at hand. The DSP chip has RAM from 250 Kb up to 2 Mb, ROM, input/output processors, DMA controllers and fast serial input/output ports. Multiprocessing system can be organized. SHARC is very compact (27 mm x 27 mm). It has low power modes that significantly reduce the power dissipation. So, several such DSP can be fixed on PCI cards. Unlike Intel Pentium it does not need a cooler. Its predecessor AD21150 is used in Monster Sound 3D cards. Today SHARC is mostly employed in powerful studio hardware (for instance, DSP/FX cards, http://www.oocities.org/SiliconValley/Pines/7899/studio/pcstudio.htm).

More specific sound processing devices are produced by Aueral and E-mu. Vortex 2, Aueral AU8820, 330 MIPS by Aueral are used in Turtle Beach Montego A3DXStream sound cards, EMU101k (1000 MIPS!) by E-mu - in SoundBlaster Live by Creative Labs. By the way on http://www3.techstocks.com/~wsapi/investor/repliestothis-3560864 you can observe funny rumors of Creative Labs efforts to copyright even most ordinary decisions (such as 8-point interpolation, data cash or table synthesis). AU8820 and EMU101k instructions are more understandable than AD21160. They enable table synthesis of musical sounds, reverb, flanger, chorus, various filters and 3D sound effects.

Vortex 2 chip supports A3D technology copyrighted by Aueral and DirectSound3D as well. It processes up to 32 3D static sound sources on sampling frequency up to 48 kHz. Aureal Wavetracing™ provides reverb effects. However the reverberation is flat judging from the picture in Aueral web-site. Say, software by Waves (http://www.waves.com) gives a similar reverb.

EMU101k chip seems to me more effective. It processes up to 148 3D sound sources using DirectSound3D technology and up to 32 sound sources on sampling frequency up to 48 kHz applying A3D. EMU101k enables reverb, chorus, equalizer and various filters effects. Unlike Aueral Creative Labs has copyrighted Environmental Audio technology. Though it seems they have bet on Microsoft DirectSound 3D.

It is interesting to witness a competition between Creative Labs and Aueral. It is for the first time that Creative Labs has a deserving rival in the market of cheap 3D sound cards. Aueral has even managed to limit Microsoft DirectSound3D influence. Aueral declares that all the leading producers of computer games (more than 100 all in all) support A3D technology. While it is practically identical to DirectSound3D.

DSPs by Motorola and Texas Instruments enable 3D sound processing as well. So the market offers numerous reliable devises. It makes possible to design relatively cheap sound cards supplying with DSP. Such sound cards can realize biphonic processors and digital filters required for quality modeling of a human sense of hearing. Thus, 3D sound has become a standard function of a modern sound card. In addition to digital filters such sound cards can perform Dopler effect which is so essential in computer games. Some sound cards with DSP (http://www.sblive.com) have a rich set of sound effects, reverbs and filters. So they can refine a CD sound considerably (though it is a matter of taste). The enthusiasts of home computer sound recording are welcome to experiment with 3D sound. All popular sound editing software have special plug-ins providing the creation of 3D sound compositions (http://www.qsound.com).

Simanenkov Dmitry (E-mail mailto:digital_sound@bigfoot.com)