This appendix covers binaural recording with an artificial (dummy) head. The head contains a microphone flush-mounted in each ear. You record with these microphones and play back the recording over headphones. This process can re-create the locations of the original performers and their acoustic environment with exciting realism.
You can substitute your own head for the artificial head by placing miniature condenser microphones in your ears, or on your temples, and recording with them. Some podcasts are made this way.
Thanks to the popularity of MP3 players with earphones, many people have the opportunity to hear binaural recordings.
Binaural Recording and the Artificial Head
Binaural (two-ear) recording starts with an artificial head or dummy head. This is a model of a human head with a flush-mounted microphone in each ear ( Figure D-1 ). These microphones capture the sound arriving at each ear. The microphones’ signals are recorded. When this recording is played back over headphones, your ears hear the signals that originally appeared at the dummy head’s ears ( Figure D-2 ). That is, the original sound at each ear is reproduced (Geil, 1979; Genuit and Bray, 1989; Peus, 1989; Sunier, 1989a, 1989b, 1989c).
Binaural recording works on the following premise. When we listen to a natural sound source in any direction, the input to our ears is just two one-dimensional signals: the sound pressures at the eardrums. If we can re-create the same pressures at the listener’s eardrums as would have occurred “live,” we can reproduce the original listening experience, including directional information and reverberation (Moller, 1989).
Binaural recording with headphone playback is the most spatially accurate method now known. The re-creation of sound-source locations and room ambience is startling. Often, sounds can be reproduced all around your head—in front, behind, above, below, and so on. You may be fooled into thinking that you’re hearing a real instrument playing in your listening room. To see some articles on binaural recording, Google “binaural John Sunier.” Sunier is a binaural expert. Samples can be heard here: https://soundcloud.com/groups/binaural-recording.
As for drawbacks: the artificial head is conspicuous, which limits its use for recording live concerts; it is not mono-compatible; and it is relatively expensive. Some sources for dummy heads are given in Chapter 12 under the heading “Dummy Heads and Headworn Binaural Mics.”
How It Works
An artificial head picks up sound as a human head does. The head is an obstacle to sound waves at middle to high frequencies. On the side of the head away from the sound source, the ear is in a sonic shadow: the head blocks high frequencies. In contrast, on the side of the head toward the source, there is a pressure buildup (a rise in the frequency response) at middle to high frequencies.
The folds in the pinna (outer ear) also affect the frequency response by reflecting sounds into the ear canal. These reflections combine with the direct sound, causing phase cancellations (dips in the response) at certain frequencies.
The human eardrum is inside the ear canal, which is a resonant tube. The ear canal’s resonance does not change with sound-source direction, so the ear canal supplies no localization cues. For this reason, it is omitted in most artificial heads. Typically, the microphone diaphragm is mounted nearly flush with the head, 4 mm (0.16 in) inside the ear canal.
To summarize: the head and outer ear cause peaks and dips in the frequency response of the sound received. These peaks and dips vary with the angle of sound incidence; they vary with the sound-source location. The frequency response of an artificial head is different in different directions. In short, the head and outer ear act as a directiondependent equalizer.
Each ear picks up a different spectrum of amplitude and phase because one ear is shadowed by the head and the ears are spaced apart. These interaural differences vary with the source location around the head.
When the signals from the dummy-head microphones are reproduced over headphones, you hear the same interaural differences that the dummy head picked up. This creates the illusion of images located where the original sources were.
Physically, an artificial head is a near-coincident array using boundary microphones: the head is the boundary, and the microphones are flush-mounted in this boundary. The head and outer ears create directional patterns that vary with frequency. The head spaces the microphones about 6 1/2 inches apart. Some dummy heads include shoulders or a torso, which aids front/back localization in human listening but can degrade it in binaural recording and playback (Griesinger, 1989).
The microphones in a near-coincident array are directional at all frequencies and use no baffle between them. In contrast, the mics in an artificial head are omni at low frequencies and unidirectional at high frequencies (due to the head baffle effect).
Ideally, the artificial head is as solid as a human head, to attenuate sound passing through it (Sunier, 1989c). For example, the Head Acoustics artificial head is made of molded, dense fiberglass (Genuit and Bray, 1989). In contrast, the Sonic Studios GUY and LiteGUY artificial heads are made of absorbent Sorbothane.
As we said, you can substitute your own head for the artificial head by placing miniature condenser microphones in your ears and recording with them. The more that a dummy head and ears are shaped like your particular head and ears, the better the reproduced imaging. Thus, if you record binaurally with your own head, you might experience more precise imaging than you would if you recorded with a dummy head. This recording will have a nonflat response because of head diffraction (which I will explain later).
Core Sound (www.core-sound.com) is the world’s largest manufacturer of binaural microphones. The company offers miniature omni condenser mics that can be clipped onto eyeglass earpieces. These mics make excellent binaural recordings. Sonic Studios (www.sonicstudios.com) has a similar product, DSM (Dimensional Stereo Microphones), that are worn on the temples rather than in the ear. Based on the head-related transfer function (HRTF), DSM mics are said to provide better stereo over loudspeakers than binaural mics can provide. HRTF is the effect of the head on the frequency response and phase response of a sound coming from a particular direction.
Another substitute for a dummy head is a head-size sphere with flush-mounted microphones where the ears would be. This system, called the Kugelflachenmikrofon, was developed by Gunther Theile for improved imaging over loudspeakers (Griesinger, 1989). See Appendix C under the heading “Sphere Microphones.” Website track 17 demonstrates the stereo imaging of a sphere microphone. Listen to it over headphones as well as loudspeakers.
Some websites of commercial products are listed in Chapter 12 under the headings “Dummy Heads and Headworn Binaural Mics” and “Stereo Microphones.”
You might hear the binaural images inside your head, rather than outside. One reason has to do with head movements. When you listen to a sound source that is outside your head and move your head slightly, you hear small changes in the arrival-time differences at your ears. This is a cue to the brain that the source is outside your head. Small movements of your head help to externalize sound sources. But headphones lack this cue because the images move with your head motion.
Headphones with head-tracking sensors can make the images appear to be stationary when you turn your head, resulting in very realistic out-of- head localization. See http://www.sony.net/Products/vpt/tech/.
Another reason for in-head localization is that the conch resonance of the pinna is disturbed by most headphones. The conch is the large cavity in the pinna just outside the ear canal. If you equalize the headphone signal to restore the conch resonance, you hear images somewhat outside the head (Cooper and Bauck, 1989).
An artificial head (or a human head) has a nonflat frequency response due to the head’s diffraction, the disturbance of a sound field by an obstacle. The diffraction of the head and pinnae creates a very rough frequency response, generally with a big peak around 3 kHz for frontal sounds. Therefore, binaural recordings sound tonally colored unless custom equalization is used. Some artificial heads have built-in equalization that compensates for the effect of the head.
What is the best equalization for an artificial head to make it sound tonally like a conventional flat-response microphone? Several equalization schemes have been proposed:
• Diffuse-field equalization: This compensates for the head’s average response to sounds arriving from all directions (such as reverberation in a concert hall).
• Frontal free-field equalization: This compensates for the head’s response to a sound source directly in front, in anechoic conditions.
• 10 ° averaged, free-field equalization: This compensates for the head’s response to a sound source in anechoic conditions, averaged over ± 10° off-center.
• Free field with source at ± 30 ° equalization: This compensates for the head’s response to a sound source 30° off-center, in anechoic conditions. This is a typical stereo loudspeaker location.
The Neumann KU-100 and KEMAR artificial heads use diffuse-field equalization, which Theile also recommends. However, Griesinger (1989) found that the Neumann head needed additional equalization to sound like a Calrec SoundField microphone: approximately 7 dB at 3 kHz and 4 dB at 15 kHz. He prefers either this equalization or a 10° averaged free-field response for artificial heads. The Head Acoustics head, developed by Gierlich and Genuit, is equalized flat for free-field sounds in front (Genuit and Bray, 1989), while Cooper and Bauck (1989) recommend that artificial heads be equalized flat for free-field sounds at ± 30°.
To provide a net flat response from microphone to listener, the artificial-head equalization should be the inverse of the headphone frequency response. If the dummy head is equalized with a dip around 3 kHz to yield a net flat response, the headphones should have a mirror-image peak around 3 kHz (most do).
Artificial-Head Imaging with Loudspeakers
How does an artificial-head recording sound when reproduced over loudspeakers? According to Griesinger (1989), it can sound just as good as an ordinary stereo recording, with superior reproduction of location, height, depth, and hall ambience. But it sounds even better over headphones.Images in binaural recordings are mainly up front when you listen with speakers but are all around when you listen with headphones.
Genuit and Bray (1989) report that more reverberation is heard over speakers than over headphones, due to a phenomenon called binaural reverberance suppression. For this reason, it is important to monitor artificial-head recordings with headphones and speakers.
Griesinger notes that a dummy head must be placed relatively close to the musical ensemble to yield an adequate ratio of direct-to-reverberant sound over loudspeakers. This placement yields exaggerated stereo separation with a hole in the middle. However, the center image can be made more solid by boosting in the presence range (see Griesinger’s, 1989, recommended equalization previously).
Although a dummy-head binaural recording can provide excellent imaging over headphones, it produces inadequate spaciousness at low frequencies over loudspeakers (Huggonet and Jouhaneau, 1987) unless spatial equalization is used (Griesinger, 1989). Spatial equalization was discussed in Appendix B under the heading “Coincident Systems with Spatial Equalization (Shuffler Circuit).” A low-frequency boost in the L − R difference signal of about 15 dB at 40 Hz and 1 dB at 400 Hz can improve the low-frequency separation over speakers.
Excerpt from Recording Music on Location, 2nd Edition by Bruce Bartlett and Jenny Bartlett © 2014 Taylor & Francis Group. All Rights Reserved.