tomwalters@0: .TH GENNAP 1 "8 April 1994" tomwalters@0: .LP tomwalters@0: .SH NAME tomwalters@0: .LP tomwalters@0: gennap \- generate neural activity pattern tomwalters@0: .LP tomwalters@0: .SH SYNOPSIS tomwalters@0: .LP tomwalters@0: gennap [ option=value | -option ] [ filename ] tomwalters@0: .LP tomwalters@0: .SH DESCRIPTION tomwalters@0: .LP tomwalters@0: The gennap module of the AIM software converts an input wave into a tomwalters@0: simulated neural activity pattern (NAP), which is AIM's representation tomwalters@0: of the pattern of information in the auditory nerve at about the level tomwalters@0: of the cochlear nucleus. Gennap begins by calculating the basilar tomwalters@0: membrane motion (BMM) associated with the input wave using the genbmm tomwalters@0: module, and then it applies several additional transforms that we know tomwalters@0: occur in some form during the neural transduction process. AIM tomwalters@0: provides two alternative methods for generating the NAP, a tomwalters@0: two-dimensional adaptive thresholding mechanism (Holdsworth and tomwalters@0: Patterson, 1993), and an array of inner haircell simulators based tomwalters@0: (Meddis et al., 1990; Giguere and Woodland, 1994). The adaptive tomwalters@0: thresholding mechanism applies rectification, log compression, tomwalters@0: adaptation in time, and suppression across frequency; its purpose is tomwalters@0: to stabilise the level of the membrane activity with compression and tomwalters@0: then sharpen the features that appear in the compressed membrane tomwalters@0: motion. Together, the gammatone filterbank and adaptive thresholding tomwalters@0: form a 'functional' cochlea simulation. The Meddis module applies tomwalters@0: level-dependant compression and adaptation that simulate the response tomwalters@0: of inner haircells to membrane motion. The cells are not coupled and tomwalters@0: so there is no frequency sharpening in this module. Together, the tomwalters@0: transmission-line filterbank and the Meddis module form a tomwalters@0: 'physiological' cochlea simulation. tomwalters@0: .LP tomwalters@0: .SH OPTIONS tomwalters@0: .LP tomwalters@0: The options for gennap are grouped according to the functions they tomwalters@0: control. The adaptive thresholding options are identified by the tomwalters@0: common suffix _at; the Meddis module options are identified by the tomwalters@0: common suffix _med. These two groups of options are the subject of tomwalters@0: this manual entry, together with two additional options that specify tomwalters@0: whether rectification and compression operations are required before tomwalters@0: the transduction stage. There is also an option to specify the choice tomwalters@0: of the transduction function. tomwalters@0: .LP tomwalters@0: .SH RECTIFICATION AND COMPRESSION tomwalters@0: .LP tomwalters@0: The adaptive thresholding process begins with rectification and log tomwalters@0: compression of the BMM. It is occasionally useful to have these tomwalters@0: functions available separately and so the options 'rectify' and tomwalters@0: 'compress' are presented separately in the options list before the tomwalters@0: neural transduction options. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: rectify tomwalters@0: Rectification switch tomwalters@0: .RS tomwalters@0: Switch. Default value: off. tomwalters@0: .RE tomwalters@0: .RS tomwalters@0: .LP tomwalters@0: If rectify is on, the BMM is half-wave rectified. tomwalters@0: The compression operation also performs half-wave rectification (to tomwalters@0: avoid taking logs of negative numbers). So the rectify option is tomwalters@0: really here just to provide for rectified BMM in the absence of tomwalters@0: compression. As a result, the default for option rectify is tomwalters@0: off. (Note: Full wave rectification is produced if rectify is set to tomwalters@0: 2. This is useful when calculating envelopes with genasa.) tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: compress tomwalters@0: Compression switch tomwalters@0: .RS tomwalters@0: Switch. Default value: on. tomwalters@0: .RE tomwalters@0: .RS tomwalters@0: .LP tomwalters@0: The compressor is strictly logarithmic and so to this point, the tomwalters@0: functional cochlea simulation is level independent. In the auditory tomwalters@0: system, the compressor is logarithmic over the lower part of its range tomwalters@0: and then it asymptotes to a soft limit. The default for option tomwalters@0: compress is on (note that the compressor also performs half-wave tomwalters@0: rectification). tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: Important: The default value for option compress is 'on' which assumes tomwalters@0: that the transduction module is adaptive thresholding (the default for tomwalters@0: the transduction option described below). If the Meddis transduction tomwalters@0: module is selected (transduction=med), compress should be set to 'off' tomwalters@0: to obtain the operation described in Giguerre and Woodland tomwalters@0: (1994). This can be done on the command line (see EXAMPLES) or in the tomwalters@0: appropriate .gen???rc files. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .SH NEURAL TRANSDUCTION tomwalters@0: .LP tomwalters@0: The neural transduction is performed either by two-dimensional tomwalters@0: adaptive thresholding or an array of Meddis haircells. The choice is tomwalters@0: controlled by the option 'transduction'. tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: transduction tomwalters@0: The transduction function tomwalters@0: .RS tomwalters@0: Switch. Default value: at. Choices: at, med, off. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: If adaptive thresholding is specified (at), the options with suffix tomwalters@0: _at below apply; if the Meddis module is specified (med), the options tomwalters@0: with suffix _med below apply. If off is specified, no transduction tomwalters@0: function is applied. The default is at. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .SS "Two-dimensional adaptive thresholding: _at " tomwalters@0: .PP tomwalters@0: The adaptive thresholding mechanism is a functional model of neural tomwalters@0: encoding. Its purpose is to enhance the contrast of the larger tomwalters@0: features that appear in the surface of the BMM and reduce those tomwalters@0: aspects of the representation which are just a direct consequence of tomwalters@0: the filtering and compression processes (Holdsworth and Patterson, tomwalters@0: 1993). The process begins with rectification and compression of the tomwalters@0: BMM. The tail of the envelope of the impulse response of the tomwalters@0: gammatone filter is exponential. As a result, logarithmic compression tomwalters@0: is used, since this makes the filter decay function linear in NAP tomwalters@0: coordinates. Following compression, adaptation is applied in time and tomwalters@0: suppression is applied across frequency. tomwalters@0: .LP tomwalters@0: Briefly, an adaptive threshold value is maintained for each channel tomwalters@0: and updated at the sampling rate. The new value is the largest of a) tomwalters@0: the previous value reduced by a fast-acting temporal decay factor tomwalters@0: (t1recovery_at), b) the previous value reduced by a longer-term tomwalters@0: temporal decay factor (t2recovery_at), c) the adapted level in the tomwalters@0: channel immediately above, reduced by a frequency spread factor tomwalters@0: (frecovery_at), d) the adapted level in the channel immediately below, tomwalters@0: reduced by the same frequency spread factor, or e) a floor level that tomwalters@0: precludes the mechanism listening to its own internal noise tomwalters@0: (reclimit_at). The mechanism produces output whenever the input tomwalters@0: exceeds the adaptive threshold, and the output level is the difference tomwalters@0: between the input and the adaptive threshold. The adaptation and tomwalters@0: suppression are coupled, and they jointly sharpen features like vowel tomwalters@0: formants which appear smeared in compressed BMM. tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: trise_at tomwalters@0: Threshold rise rate tomwalters@0: .RS tomwalters@0: Default value: 1000. tomwalters@0: .RE tomwalters@0: .RS tomwalters@0: .LP tomwalters@0: Upward Adaptation: This option specifies the rate at which the tomwalters@0: adaptive threshold will rise in response to a rise in signal tomwalters@0: level. The default value, 1000, means that the adaptive threshold tomwalters@0: responds very quickly to increases in the input wave; essentially, it tomwalters@0: follows the envelope of any rise in signal amplitude. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: Downward Adaptation: Following the cessation of sound, or a rapid drop tomwalters@0: in input level, temporal adaptation occurs in two stages as determined tomwalters@0: by t1recovery_at, t2recovery_at and propt2t1_at: If the default values tomwalters@0: are used, the mechanism initially adapts at a rate slightly slower tomwalters@0: than the decay rate of the gammatone filter in the given channel, and tomwalters@0: this represses much of the ringing of the impulse response of the tomwalters@0: filter. Later the adaptation switches to a slower rate more in line tomwalters@0: with data on auditory forward masking. The option propt2t1_at tomwalters@0: determines the point at which the initial fast rate of decay gives way tomwalters@0: to the slower limiting decay rate. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: t1recovery_at tomwalters@0: The initial rate of threshold recovery relative to filter decay rate tomwalters@0: .RS tomwalters@0: Default value: 0.6. tomwalters@0: .RE tomwalters@0: .RS tomwalters@0: .LP tomwalters@0: This option determines the initial rate of decay of the adaptive tomwalters@0: threshold relative to the rate of decay of the auditory filter, tomwalters@0: provided propt2t1_at is less than unity. Values of t1recovery_at less tomwalters@0: than unity cause the adaptive threshold to decay more slowly than the tomwalters@0: auditory filter and thereby to remove the filter response from the tomwalters@0: representation when it is the sole reason for BMM activity. The rate tomwalters@0: of decay is linear with respect to the log-compressed BMM, so it is tomwalters@0: like an exponential decay with respect to the BMM. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: t2recovery_at tomwalters@0: The secondary threshold recovery rate tomwalters@0: .RS tomwalters@0: Default value: 0.2. tomwalters@0: .RE tomwalters@0: .RS tomwalters@0: .LP tomwalters@0: This option determines the limiting rate of decay of the adaptive tomwalters@0: threshold when the sound ceases provided propt2t1_at is less than tomwalters@0: unity. The default value causes the adaptive threshold to decay more tomwalters@0: slowly than the initial rate as observed in auditory forward masking. tomwalters@0: Note, however, that the system to this point is level independent, tomwalters@0: whereas auditory forward masking is level dependent. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: propt2t1_at tomwalters@0: The point at which t1recovery_at gives way to t2_recovery_at tomwalters@0: .RS tomwalters@0: Default value: 0.5. tomwalters@0: .RE tomwalters@0: .RS tomwalters@0: .LP tomwalters@0: This option determines the point at which the initial fast rate of tomwalters@0: decay (t1recovery_at) gives way to the slower limiting decay rate tomwalters@0: (t2recovery_at). The nomanclature assumes that propt2t1_at is a value tomwalters@0: less than unity. Otherwise the the roles of the initial and limiting tomwalters@0: decays are reversed. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: frecovery_at tomwalters@0: Recovery rate across frequency tomwalters@0: .RS tomwalters@0: Default value: 20. tomwalters@0: .RE tomwalters@0: .RS tomwalters@0: .LP tomwalters@0: This parameter specifies the rate at which a threshold value in one channel tomwalters@0: propagates to influence threshold in neighbouring channels. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: reclimit_at tomwalters@0: Limitation on recovery level tomwalters@0: .RS tomwalters@0: Default units: mB. Default value: 500 mB. (mB=milliBells) tomwalters@0: .RE tomwalters@0: .RS tomwalters@0: .LP tomwalters@0: In order to prevent the mechanism from encountering system noise, tomwalters@0: or alternately, to reduce sensitivity to stimulus noise, there is a tomwalters@0: limit placed on the recovery that the adaptive threshold can achieve. tomwalters@0: The limit, reclimit_at, is the limit of the sensitivity of the system. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: gain_at tomwalters@0: Output gain tomwalters@0: .RS tomwalters@0: Default units: scalar. Default value: 1. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .SS "Meddis haircell model: _med " tomwalters@0: .PP tomwalters@0: The purpose of the Meddis module is to simulate neural transduction of tomwalters@0: BMM as performed by the inner haircells of the cochlea. There is one tomwalters@0: haircell simulation unit for each output channel of the filterbank. tomwalters@0: The haircell equations (Meddis et al., 1990) are solved using the wave tomwalters@0: digital filter algorithm described in Giguere and Woodland (1994). The tomwalters@0: characteristics of the haircell are controlled by options: fiber_med, tomwalters@0: thresh_med, and gain_med. tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: fiber_med tomwalters@0: The spontaneous-rate of the simulated fiber tomwalters@0: .RS tomwalters@0: Default value: medium. Choices: medium, high. tomwalters@0: .RE tomwalters@0: .RS tomwalters@0: .LP tomwalters@0: If medium is specified, a medium spontaneous-rate haircell fiber is tomwalters@0: simulated. If high is specified, a high spontaneous-rate tomwalters@0: fiber is simulated. The properties of these two types of fibers tomwalters@0: are listed in Table II in Meddis et al. (1990). tomwalters@0: The default value is medium. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: thresh_med tomwalters@0: The threshold shift of the fiber tomwalters@0: .RS tomwalters@0: Default Units: dB. Default value: 0. tomwalters@0: .RE tomwalters@0: .RS tomwalters@0: .LP tomwalters@0: This option shifts the entire rate-intensity function of the haircell tomwalters@0: fiber horizontally to a higher or lower level, to accomodate changes tomwalters@0: in the scaling of the input wave. A positive (negative) value tomwalters@0: increases (decreases) the rate- and saturation-thresholds of the fiber tomwalters@0: by that amount. This operation does not change the dynamic range, the tomwalters@0: spontaneous and saturation rates, or the adaptation time constants or tomwalters@0: synchronization index of the fiber. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: gain_med tomwalters@0: Output gain tomwalters@0: .RS tomwalters@0: Default units: scalar. Default value: 1. tomwalters@0: .RE tomwalters@0: .RS tomwalters@0: .LP tomwalters@0: Note: There is an internal gain of 20.0 within the software of tomwalters@0: the Meddis haircell model itself. The total gain is therefore tomwalters@0: 20.0 times the value for gain_med. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .SH REFERENCES tomwalters@0: .LP tomwalters@0: .RE tomwalters@0: .TP 4 tomwalters@0: Giguere, C. and Woodland, P.C. (1994). A computational model of tomwalters@0: the auditory periphery for speech and hearing research. I. Ascending tomwalters@0: path. J.Acoust. Soc. Am. 95: 331-342. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 4 tomwalters@0: Holdsworth, J. (1990). Two-Dimensional adaptive thresholding. tomwalters@0: Annex 4 of AAM-HAP Report 1, APU contract Report. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .TP 4 tomwalters@0: Holdsworth, J. and Patterson, R.D. (1993). "Analysis of waveforms," tomwalters@0: UK Patent GB 2234078B. tomwalters@0: .LP tomwalters@0: .TP 4 tomwalters@0: Meddis, R., Hewitt, M. and Shackleton, T. (1990). Implementation tomwalters@0: details of a computational model of the inner-haircell/auditory-nerve tomwalters@0: synapse. J.Acoust. Soc. Am. 87: 1813-1816. tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: .SH EXAMPLES tomwalters@0: .LP tomwalters@0: The following command generates the neural activity pattern using the tomwalters@0: gammatone auditory filterbank (the default) and the adaptive tomwalters@0: thresholding (the default) for an input file named cegc: tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: example% gennap cegc tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: The following command generates the neural activity pattern using the tomwalters@0: gammatone filterbank (the default) and Meddis haircell tomwalters@0: transduction for input cegc: tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: example% gennap compress=off transduction=meddis cegc tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: The following command generates the neural activity pattern using the tomwalters@0: transmission line filterbank and Meddis haircell transduction for cegc: tomwalters@0: .RE tomwalters@0: .LP tomwalters@0: example% gennap filter=tlf compress=off transduction=meddis cegc tomwalters@0: .LP tomwalters@0: .SH FILES tomwalters@0: .LP tomwalters@0: .TP 13 tomwalters@0: .gennaprc tomwalters@0: The options file for gennap. tomwalters@0: .LP tomwalters@0: .SH SEE ALSO tomwalters@0: .LP tomwalters@0: genepn, gencgm, genbmm tomwalters@0: .LP tomwalters@0: .SH COPYRIGHT tomwalters@0: .LP tomwalters@0: Copyright (c) Applied Psychology Unit, Medical Research Council, 1995 tomwalters@0: .LP tomwalters@0: Permission to use, copy, modify, and distribute this software without fee tomwalters@0: is hereby granted for research purposes, provided that this copyright tomwalters@0: notice appears in all copies and in all supporting documentation, and that tomwalters@0: the software is not redistributed for any fee (except for a nominal tomwalters@0: shipping charge). Anyone wanting to incorporate all or part of this tomwalters@0: software in a commercial product must obtain a license from the Medical tomwalters@0: Research Council. tomwalters@0: .LP tomwalters@0: The MRC makes no representations about the suitability of this tomwalters@0: software for any purpose. It is provided "as is" without express or tomwalters@0: implied warranty. tomwalters@0: .LP tomwalters@0: THE MRC DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING tomwalters@0: ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL tomwalters@0: THE A.P.U. BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES tomwalters@0: OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, tomwalters@0: WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, tomwalters@0: ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS tomwalters@0: SOFTWARE. tomwalters@0: .LP tomwalters@0: .SH ACKNOWLEDGEMENTS tomwalters@0: .LP tomwalters@0: The AIM software was developed for Unix workstations by John tomwalters@0: Holdsworth and Mike Allerhand of the MRC APU, under the direction of tomwalters@0: Roy Patterson. The physiological version of AIM was developed by tomwalters@0: Christian Giguere. The options handler is by Paul Manson. The revised tomwalters@0: SAI module is by Jay Datta. Michael Akeroyd extended the postscript tomwalters@0: facilites and developed the xreview routine for auditory image tomwalters@0: cartoons. tomwalters@0: .LP tomwalters@0: The project was supported by the MRC and grants from the U.K. Defense tomwalters@0: Research Agency, Farnborough (Research Contract 2239); the EEC Esprit tomwalters@0: BR Porgramme, Project ACTS (3207); and the U.K. Hearing Research Trust. tomwalters@0: