Binaural sound recording

from Wikipedia, the free encyclopedia

A binaural sound recording is a sound recording of sound signals with microphones which, when played back only via headphones , are intended to produce a natural auditory impression with precise directional localization. An artificial head is often used when recording in artificial head stereophony .

Stereo and binaural

The term "binaural" (etymologically derived from "binary aurality", ie hearing with both ears ) was often equated with stereo in the past .

In general, stereo sound recordings are mixed solely via loudspeaker systems when listening, hence the name "loudspeaker stereophony". The properties used by humans for localization such as the shape of the head or auricle are rightly not taken into account; because with natural hearing and playback via stereo loudspeakers in the stereo triangle , the hearing itself generates the ear signals .

Binaural recordings are "stereo" recordings with special recording technology that are typically only reproduced correctly with headphones, hence the designation as "headphone stereophony". Binaural recordings, which replace the natural ear signals suppressed by headphones, are the best technical solution for realistically reproducing a spatial hearing impression.

Basics of hearing

Definition of an audio event

A sound event only becomes an auditory event when the sound waves have penetrated the ear, are present in the brain as a stimulus and have been processed by it into a hearing perception. Sound and auditory events cannot be compared directly either, since the shape of the stimuli is changed in the middle and inner ear. This means that the ear somewhat distorts the signal. These stimuli are sensations that differ from person to person and depend on the frequency , duration and sound pressure level of the sound event. Sensations cannot be measured, but can be statistically recorded psychoacoustically through listening studies .

Localization of sound events

Designation of the three levels
above: 1st horizontal level (transversal level)
 middle: 2nd median level (sagittal level)
below: 3rd frontal level

Humans are able to assign their perceived auditory events to specific directions. Humans have mechanisms of perception for half the horizontal plane and the median plane . In a head-related coordinate system, the 0 ° axis is defined as the horizontal forward direction.

Horizontal plane

As soon as the sound source is no longer directly on the 0 ° axis, identical oscillation phases reach the left and right ears at different transit times and at different levels ( interaural transit time differences , ITD and interaural level differences , ILD ) and induce the hearing to locate a room in virtual sound field. The differences here are tiny: time differences (time differences) can be evaluated by the human ear from a size of 10 µs for directional localization. This corresponds to a localization sharpness in the horizontal plane of about one degree. Up to a transit time difference of 0.63 ms (i.e. 630 µs), the lateral localization increases roughly proportionally to the transit time difference. A transit time difference of 0.63 ms corresponds to a distance difference of the sound of 21.5 cm. This variable, also known as the “Hornbostel-Wertheimer constant”, corresponds to the difference in distance when sound is incident from 90 ° or 270 °, with an average distance between the two ears on the human head.

Sound can be localized in the horizontal plane through interaural signal differences. Here, however, it is not possible to differentiate between sound from the front and rear horizontal plane, because the interaural signal differences are the same. A distinction is made between three frequency ranges; see also duplex theory by John William Strutt, 3rd Baron Rayleigh .

  • Below 800 Hz, the determination of interaural transit time differences (ITD) is based primarily on the evaluation of phase differences between the ear signals. Interaural level differences play no role here.
  • Above 1600 Hz, the localization is based on the evaluation of interaural level differences (ILD) as well as on the evaluation of interaural group delay differences (delay differences of the signal envelopes).
  • In the range between 800 Hz and 1600 Hz, the effective ranges of the effects involved overlap. With increasing frequency, the angular range in which interaural phase differences can be evaluated becomes smaller and smaller. Instead, the size of the interaural level differences increases.

Interaural level differences arise from shadowing by the head. Interaural level differences are strongly frequency dependent. Frequencies with wavelengths in the order of magnitude of the obstacle are hardly deflected. Low frequencies below about 300 Hz do not form a sound shadow and therefore no level differences that can be used to localize the direction. However, these low frequencies with the phase differences have a special influence on the sense of space and the feeling of being enveloped.

Median plane

In the median plane there are almost no time differences and level differences between the two ears. The acoustic properties of the outer ear are used here to localize a sound event .

The different elevations and depressions of the auricle , together with the auditory canal, form an acoustic resonator system that is stimulated differently depending on whether the sound comes from the front, from above or from behind. This results in direction-dependent minima and maxima in the frequency response of the ear sensitivity. The sound changed as a result is localized by the hearing in the median plane and corrected for sound.

The frequency ranges that are excited by certain directions of incidence are also called direction-determining bands . Since the shape of the outer ear is different in every person, every person also has slightly different "frequency responses" for the front, top and back.

Sound events recorded with the artificial head that lie in the median plane are often difficult to localize when listening to them later. This often leads to front-to-back or top-to-bottom reversals or in-head localization when playing headphones. One reason is that the artificial head does not generate the correct frequency structure because its spatial filter effect does not match the individual head-related transfer function of the listener.

Naturally, an auditory event for this type of localization should not be in a frequency band that is too narrow. And accordingly the use of an equalizer is counterproductive here. Otherwise, by manipulating the frequency response, direction-determining frequency bands may also be changed, which then increasingly leads to localization errors (localization errors).

The localization sharpness is around 17 degrees for sound events that are unknown to the listener, and around 9 degrees for everyday sound events. These values ​​apply when the head is in a straight position. The further a signal emerges from the forward direction, the worse the localization accuracy becomes.

In the head localization

The in-head localization is an effect that is perceived as unpleasant and can occur especially when listening to headphones, but also when playing back from speakers. The auditory events are then no longer localized outside the head, but in the head.

The brain compares the incoming ear signals for known interaural signal differences and known spectral structures. If the signals have interaural time and level differences that are unknown in this combination, or if the signals have a spectral characteristic that has never been perceived with the own outer ear, localization in the head can occur. With loudspeakers, this can happen if the polarity is reversed and if they are set up at an angle of more than 90 degrees.

Such a localization can be simulated experimentally by mirroring ( inverting ) the amplitude values ​​of one of the two channels in the case of a stereo signal whose two channels are identical and listening to the resulting signal with headphones.

Diffuse field equalization and free field equalization

In the early days of this recording process, all artificial heads were free-field equalized. At the beginning of the 1980s, the Institute for Radio Technology and the Neumann company developed the KU80 into the KU81, the only difference being the equalization of the microphones. The KU80 was not suitable for reproducing the artificial head recordings via loudspeakers, which is why the free-field equalization used at the time was questioned.

Measurements in the free sound field ( free field ) are carried out without reflecting acoustic boundary surfaces. These conditions are obtained with a good approximation in an anechoic room. The resulting free field frequency response only applies to a certain direction of sound incidence . Since a flat frequency response is desired, the signal must be equalized with the help of a filter. In the case of pressure microphones, such as those used in an artificial head, the diffuse field frequency response is very different from the free field frequency response. This is due to the fact that with pressure microphones with direct sound from the 0 ° direction of discussion, with a diaphragm diameter of around 18 mm, the level of transmission is increased by 6 dB at 10 kHz. This is caused by the sound waves, the wavelength of which corresponds to the membrane diameter or is smaller. These are reflected on the membrane and the sound pressure doubles on the membrane. In the diffuse field, this leads to a drop in height, since frequencies with a smaller wavelength are no longer bent around the microphone capsule. However, this only affects frequencies from the side or the rear of the sound. For sound waves in a certain height range that impinge on the membrane from the front, i.e. from the vicinity of the front direction of sound incidence, there is a level increase of 6 dB. Since the artificial head is not intended for recording in the near field , but rather has a greater distance from the sound source, the diffuse field frequency response plays a significantly greater role. The diffuse field is characterized by evenly incident sound components from all directions. In the diffuse sound field there is not only one direction of sound incidence, as is the case with the free sound field.

Transmission dimensions of a pressure microphone

When playing with headphones, headphones with diffuse field equalization should also be used. Headphones with a flat diffuse field transmission factor offer optimal sound neutrality. It would be better to use headphones with special IRT equalization.

Binaural recording technology

Dummy head with external microphones

Without artificial head

The simplest binaural recording method requires two microphones that point sideways away from each other and have a distance ( microphone base ) of about 17 cm to 22 cm from one another. 17.5 cm, which approximately represent the position of the ear canals of an average person, are popular. A separating body that absorbs or reflects the sound, such as a soccer ball or a metal plate, is placed between the microphones in order to approximate a head. A well-known version of this arrangement is the Jecklin disk OSS, an absorbing, 30 cm disk between two microphones with omnidirectional characteristics with a microphone base of 16.5 cm. The microphones are turned slightly outwards. In Jecklin's scripts one can recently find the enlarged disc diameter of 35 cm and the now more than doubled microphone base of 36 cm.

This has not yet been generally noticed, as the outdated 30 cm disc dimensions and the too small 16.5 cm microphone base are still used.

Another head-like arrangement is the spherical surface microphone .

Binaural dummy head recordings

More complex techniques consist of exact head replicas using an artificial head. A typical binaural recording device has two condenser studio microphones with omnidirectional characteristics , which are inserted in the ear canal of the artificial head. Here, the head-related transfer functions HRTF developed in psychoacoustic research communities are reproduced.

The first stereo artificial head with a replica of the human ear canal was built in 1933. The Neumann KU-81 and KU-100 artificial heads are the most widely used binaural microphones today. The Kemar system (registered name: KEMAR) is an alternative. The more complex HMS system from Head Acoustics (registered name: HEAD acoustics) from Herzogenrath near Aachen has an automatic frequency response setting and provides a better all-round impression. It is mainly used in acoustic measurement technology.

The first radio play in artificial head stereophony broadcast on German radio was the RIAS / BR / WDR production “Demolition” (The demolished man) based on the novel by Alfred Bester for the 1973 radio exhibition in Berlin .

Further binaural microphone system alternatives

There are also alternatives, such as the ear stud with microphone, that work in a similar way.

These "earplugs with microphone", also known as the "original head microphone" (OKM), look like normal headphones, as we know them from a Walkman . They are provided with electret microphone capsules with omnidirectional characteristics. The microphones are equipped with 3.5 mm jack plugs and are therefore suitable for mobile use, as they can be recorded with standard MiniDisc players or DAT recorders .

You can take advantage of the disadvantage that you are always tied to a human head that is certainly never kept very still. Choreographies are possible for radio plays that would not be easy to produce with a conventional artificial head. So you can not only make movements around the artificial head, but also consciously include the artificial head in the piece. Many bootlegs of live concerts are produced with the help of these microphones, since the recording earplugs are very inconspicuous to wear and can provide useful results for headphone playback.

The playback

A binaural sound recording produced in this way can only be experienced optimally “spatially” when using headphones. The reproduction via loudspeakers does not give this impression, but only a somewhat "hollow" sounding stereo effect. The attempt to explain artificial head recordings with a diffuse field equalized artificial head also compatible for loudspeaker reproduction was not accepted due to aural weaknesses.

Listening with headphones results in a listening experience that can exceed the spatiality of the usual loudspeaker sound image, as it enables a more precise binaural mapping of the sound waves . Although the right-left localization is reliable, the identification of the up-down position of audio signals is more difficult. The localization of a frontal sound event also causes problems, since it appears to be shifted in height as an elevation at a certain angle . It is also typical that the signals actually present in the front are often heard from behind and less often the other way around, so that one cannot honestly speak of a front-to-back exchange.

Loudspeaker playback

Speaker reproduction is one of the biggest problems with artificial head recording. When reproducing the signals, care should be taken that both signals, i.e. that of the right side and that of the left side, arrive completely separately at the two ears. More precisely: The left ear may only receive signals from the left stereo channel and the right ear only signals from the right channel. This is of course the case with headphones, but not with a normal stereo setup of the speakers. Furthermore, the reverberation times of the listening room are added to those already recorded. There is not much that can be done about this.

Loudspeaker installation of the Heinrich Hertz Institute in Berlin and the Institute for Technical Acoustics of the TU Berlin

The Heinrich Hertz Institute in Berlin and the Institute for Technical Acoustics at the TU Berlin have proposed a solution to the problem of channel separation . As can be seen in the sketch, four loudspeakers are set up here. The speakers on the opposite side “work together”. This means that the rear speakers are each phase rotated, processed with a filter (treble reduction) and delayed in each case in such a way that they are canceled. It is necessary to maintain the correct listening position with this speaker setup.

The second proposed solution comes from the 3rd Physics Institute in Göttingen. Here each ear is covered with a loudspeaker. The usual stereo setup of the loudspeakers is no longer used here, but the loudspeakers are positioned next to the listener. There are level and transit time differences in the ears, and the frequency response is also changed. All frequencies that are in the range of head size and smaller are reflected on the head and no longer bent. The signals that still arrive at the other ear are phase-shifted due to the differences in transit time and thus possibly canceled. This method also requires a firm adherence to the listening position and is also sensitive to head turning. If a listener cannot maintain the optimal listening position and sits outside, he will still get a sound transmission similar to the stereo setup, but sound colouration must be expected.

Loudspeaker installation of the 3rd physical institute in Göttingen


In principle, the artificial head can be used wherever it is a matter of getting a realistic recording of what a listener would receive at that point. This is why the artificial head is used for measurements, radio plays and, albeit rarely, for music recordings. In these different application areas, different artificial head models with different properties are also used.


Often, normal measurement microphones are used for measurements, but they are not really suitable for many applications because they do not take into account the spatial impression, i.e. also the human hearing perception, when evaluating the acoustic signals. Investigations that include human characteristics can only be carried out with the help of the artificial head. When taking measurements in industry, the artificial head is often used as a kind of “dummy” that is in the immediate vicinity of explosions, accidents, loud machines or the like. The artificial heads for measurements are usually also connected to special computer interfaces in order to evaluate the acoustic signals. Measurements that also contain psychoacoustic parameters, such as the listener's expectations, are then mostly also evaluated in connection with listening studies.

Radio plays

Especially in the 1970s, the artificial head was also used in creative radio plays. But even today you can still find a few radio play productions that were recorded with the artificial head, especially by public broadcasters. In radio plays you can clearly hear that the positioning of the dummy head alone has an impact on the plot of the piece and the mood that arises in it, which is also used deliberately.

Music productions

The artificial head technique is only used in very few music productions. This is not due to the associated costs, but to the lack of speaker compatibility. Artificial head recordings sound uncomfortable discolored through loudspeakers, because the artificial head's own spectral differences are supplemented by your own ear signals, and that double and crossover.

In the case of music recordings, a distinction must be made between the "take-by-take" method common in rock and pop music on several individual tracks and pure stereo recordings. This procedure is also possible with artificial head recordings. Before starting the recording, you have to distribute all the instruments around the artificial head as they should be placed in the sound image later. You can then record the individual instruments one after the other on a multitrack recorder as usual. During the later mixdown you just have to make sure that the individual signals should no longer be processed without restriction with an equalizer, as this affects the directional frequency bands that are important for localizing the respective instrument. Editing with the panpot should also be avoided. Restrictions in the possibilities for post-processing are the main reason why the process has not been able to gain wide acceptance in today's standard studio productions. Audio productions are now an art product that is supposed to create an appealing perception on a home system that does not necessarily have to match the original event. But artificial head stereophony remains the only way to reproduce the sound field of the recording room in all three spatial dimensions, i.e. really spatially, to this day, alongside ambisonics and WFS holophony .

See also

Sound carrier in artificial head stereophony

  • The Bremer Kulturgesellschaft Dacapo gGmbH has been producing a series of CDs with the HMS artificial head system from Head Acoustics under the direction of Ingo Ahmels since the mid-1990s on their label d'c records. These include classical recordings with the Philharmonic State Orchestra Bremen and the CD "d'c 9 - Alphorn in Manhattan ( Mike Svoboda 's Alphorn Special, episode 2)". This 1998 CD contains recordings made in the acoustic environment of New York City.
  • On the LP / CD Tales of Mystery and Imagination by " Alan Parsons Project " the thunderstorm was recorded in artificial head technology in the title The Fall of the House of Usher .
  • Edgar Froese : Aqua . LP 1974
  • Jane : Fire, Water, Earth & Air . LP 1976
  • Lou Reed : The Bells . LP 1979
  • delta-acoustic-sampler: dummy head dimensions . LP 10-130-1
  • Code III: Planet of Man . Delta-Acoustic LP 25-125-1
  • Seedog: We hope to see you . Delta-Acoustic LP 25-126-1
  • Head songs: folklore . Delta-Acoustic LP 25-127-1
  • Golem: Golem . Delta-Acoustic LP 25-127-1
  • Harlis : Night meets the day . Sky-Records 1977, new edition: Sireena Records 2009
  • old music samplers: tastings . Delta-Acoustic LP 25-129-1
  • Pearl Jam : Binaural . This CD from 2000 contains some binaural pieces.
  • Staccato special . CD 1989, hearing test examples in artificial head stereophony. The CDs Stakkato and Stakkato 2 from 1981 and 1988 also contain artificial head and artificial head-like recordings. All three CDs were distributed by "AUDIO" magazine and represent samplers of various audio samples.
  • Hanns Dieter Hüsch : Night performance . CD 1995, track 8 was recorded in artificial head stereo.
  • Ulrich Gerhardt , Klaus Krüger , Hans Ulrich Minke : Demolition . Radio play from 1973.
  • Walter Adler : Centropolis . MC 1997, radio play from 1975.
  • Don Haworth: On a day in summer in a garden . Radio play, directed by Albrecht Surkau, Rundfunk der DDR 1981 / HörZeichen Gerichshain 2000, ISBN 978-3934492011 .
  • Henry Sorg: At the Fischbach . MP3 nature recordings from 2008
  • Vernon Reid : Unborne Embrace from the CD Mistaken Identity
  • The biologist Walter Tilgner has - beginning with Waldkonzert (1985) - published a series of CDs with so-called natural audio images in artificial head stereophony.
  • The friends of green sonic label publishes music productions using the new pantophonic artificial head technology . Duo Seraphim, a Renaissance Journey (2015), CD 1428, HRA 1428 (44.1kHz / 24 bit recording).


  • On the theory of optimal reproduction of stereophonic signals via loudspeakers and headphones. Rundfunktechnische Mitteilungen (RTM), 1981 ( PDF; 1.8 MB ).
  • H. Hudde, Schröter: Improvements to the Neumann artificial head system. JRTM, 1981.
  • R. Kürer, G. Plenge, H. Wilkens: Reproduction of artificial head signals over loudspeakers. Radio Mentor, 1973.
  • Stephan Peus: Natural hearing with an artificial head . Funkschau 1983 ( PDF; 349 kB ).
  • Neumann brochure: The artificial head - theory and practice . ( PDF; 4.8 MB )
  • G. Theile: The importance of diffuse field equalization for stereophonic recording and playback. 13th Tonmeistertagung, Munich 1984, conference report, pp. 112–124.
  • R. Kohemann, K. Genuit: Use of artificial head technology in music productions .
  • Matthias Thalheim: Dramaturgically staging consequences of artificial head stereophony in funk-dramatic productions . Diploma thesis, Humboldt University of Berlin 1985, Section Cultural Studies and Aesthetics, Theater Studies, Neoepubli Verlag, Berlin 2016, ISBN 9783737597814 .
  • Johannes Webers: Recording studio technology . Franzis, 1985, ISBN 3-7723-5524-2 .

Web links

Commons : Binaural sound recording  - collection of images, videos and audio files

Individual evidence

  1. OSS technology (Jecklin disk) . ( Memento from October 19, 2013 in the Internet Archive ) PDF-Script, p. 32 at the bottom.
This version was added to the list of articles worth reading on July 26, 2006 .