About

Jont Allen is a professor at The Grainger College of Engineering at the University of Illinois Urbana-Champaign, with a Ph.D. in Electrical Engineering from The University of Pennsylvania obtained in 1970. His extensive career includes 32 years at AT&T Bell Labs, where he specialized in nonlinear cochlear modeling, auditory and cochlear speech processing, and speech perception. During his tenure at AT&T, Allen authored over 50 sole-authored journal articles on hearing, cochlear modeling, signal processing, room acoustics, speech perception, and the articulation index. He played a key role in developing the first commercial multiband wideband dynamic range compression hearing aid, later sold as the ReSound hearing aid, and developed the first DSP code and fitting system based on loudness in 1/2 octave bands. Allen also developed one of the first systems for non-evasively evaluating cochlear hearing using distortion product otoacoustic emissions, which was commercially sold by Etymotic Research and Mimosa Acoustics, where he serves as CTO. Since joining the University of Illinois in 2003, Allen has focused on noninvasive objective diagnostic testing of cochlear and middle ear function, speech processing for hearing aid applications, speech and music coding, and models of loudness and masking. His research aims to improve hearing aid signal processing and automatic speech recognition robustness in noisy environments. Allen has led numerous research projects, including large databases of speech perception in noise, and has collaborated extensively on reading disabilities in children. His work has provided deep insights into speech perception, particularly the features of plosive and fricative sounds, and has contributed to the understanding of cochlear modeling, auditory neurophysiology, and clinical audiology. Allen holds more than 20 US patents related to hearing aids, signal processing, and middle ear diagnostics, and teaches courses in mathematical physics, speech processing, signal processing, and audiology.

Research topics

• Computer Science
• Physics
• Acoustics
• Mathematical analysis
• Electrical engineering
• Mechanical engineering
• Electronic engineering
• Engineering
• Theoretical physics
• Materials science
• Geometry
• Mathematics

Selected publications

• Causal black body radiation is quantized in time and frequency

  The Journal of the Acoustical Society of America · 2025-10-01

  article1st authorCorresponding

  This research explores causal black body radiation. The introduction of the causality postulate leads to a quantized time-frequency spectrum, as apposed to the continuous spectrum, associated with Planck's classical black body thermodynamics. This quantized radiation is functionally equivalent to 1° K, as has been observed for black holes and the Josephson junction. Applying the Laplace transform to Planck’s law reveals these quantized eigen-frequencies. The quantized spectrum naturally suppresses the ultraviolet catastrophe, thus aligns with the characteristic frequencies of gravitational waves as detected by LIGO.

  PublisherDOI

• My 50 years of cochlear modeling

  AIP conference proceedings · 2024-01-01

  articleOpen access1st authorCorresponding

  The goal of this presentation is two-fold: The primary goal is to discuss my present understanding of cochlear function. A secondary goal is to review my earlier (1970-2021) cochlear modeling work, along with the roles of four close friends: Egbert De Boer, Steve Neely, Paul Fahey and George Zweig. To understanding of how the cochlea works, one needs an understanding of the experimental data on: 1) cochlear function (both basilar (BM) and tectorial membranes (TM)), 2) tympanic membrane, 3) middle ear (ME), 4) inner and outer hair cells (IHC, OHC), 5) auditory nerve (AN), and 6) cochlear amplifier (CA). My views on these topics have been greatly sharpened by looking back and unifying this complex puzzle. A great deal of progress has been made in the last 50 years. Conclusions: My recent review of neural tuning curve data from 1985, using nonlinear (NL) distortion product generation, has revealed a deeper understanding of cochlear function. The most important, and surprising result, is that the cochlea is much more linear in its filtering properties than I previously assumed. When the suppressor frequency fs is at least 1/2 octave lower than the characteristic ("best") frequency (fcf ), it is best known as "low-side" suppression. There is no "low-side" suppression for suppressors below 65 [dB-SPL] Fahey and Allen [1]. For suppressors above 65 [dB-SPL], suppression is engaged, with a slope between 1-2 [dB/dB]. Since the excitation threshold is also 65 [dB-SPL], we conclude that the neural threshold of excitation to both the inner and outer hair cells have nearly the same threshold. That is the suppression threshold of the OHC are nearly equal to, the IHC threshold. This raises the interesting question: If the IHC and OHC 65 [dB] thresholds are the same in the tails of the tuning curves, how can the CA function at threshold levels? Furthermore this is a highly unexpected result because low-side suppression, as measured on the basilar membrane, has a 20-30 [dB] higher threshold [2, 3]. Is the OHC action restricted to the neighborhood of the neuron's best frequency? This would require that the neural low-side suppression and loudness recruitment (the reduced loudness of low-intensity sounds in the hearing-impaired ear) are closely related (i.e., are the same phenomena). The ramifications of this observation seem significant as they must impact our fundamental understanding of hearing and thus hearing loss [4], (p. 332, Allen90) [5]. In summary: Two-tone suppression acts like an automatic gain control, elevating the loudness threshold, with little audible distortion. We then discuss the properties of the CA, functionally measuring the CA gain. The URL for cited manuscripts: https://auditorymodels.org/index.php?n=Main.Publications; https://www.mechanicsofhearing.org/

  PublisherOA PDFDOI

• Chaotic Convergence of Newton's Method

  IEEE Transactions on Signal Processing · 2024-01-01 · 4 citations

  articleOpen access1st authorCorresponding

  Problem statement: In 1680 Newton proposed an algorithm for finding roots of polynomials. His method has since evolved, but the core concept remains intact. The convergence of Newton’s Method has been widely challenged to be unstable or even chaotic. Here we briefly review this evolution, and consider the question of stable convergence. <p xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Methods: Newton’s method may be applied to any complex analytic function, such as polynomials. Its derivation is based on a Taylor series expansion in the Laplace frequency <inline-formula><tex-math notation="LaTeX">$s=\sigma+j\omega$</tex-math></inline-formula>. The convergence of Newton’s method depends on the <i>R</i>egion of Convergence (RoC). <p xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Findings: Under certain conditions, non-linear (NL) limit-cycles appear, resulting in a reduced rate of convergence to a root. Since Newton’s method is inherently complex analytic (that is, linear and convergent), it is important to establish the source of this NL divergence, which we show is due to violations of the Nyquist Sampling theorem, also known as aliasing. Here the conditions and method for uniform convergence are explored. <p xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Conclusions: The source of the nonlinear limit-cycle is explained in terms of aliasing. We numerically demonstrate that reducing the step-size always results in a more stable convergence. The down side is that it always results in a sub-optimal convergence. It follows that a dynamic step-size would be ideal, by slowly increasing the step-size until it fails, and then decreasing it in small steps until it converges. Finding the optimal step-size is a reasonable solution.

  PublisherOA PDFDOI

• Estimating human neural tuning curves at low-level sound from human psychophysical masking data and the cochlea-frequency maps

  The Journal of the Acoustical Society of America · 2022-10-01

  articleSenior author

  We proposed a three-slope method to estimate human frequency tuning curves at low sound levels. As a starting point, cat neural excitation patterns (Allen and Fahey, 1993) are derived from Liberman’s cat cochlea neural tuning curves data (Liberman, 1978) in a mathematical way. A three-slope structure inspired by (Rhode,1978) is proposed to represent neural excitation patterns. The aim of this paper is to estimate the three-slope structure for human neural excitation patterns and then transform human neural excitation patterns into human tuning curves by using a human cochlea-frequency map (Greenwood, 1990). In 1993, Allen and Fahey introduced the concept of the second cochlea map, which was shown to be highly correlated with both neural excitation patterns and distortion product otoacoustic emissions (DPOAE). Thus, three slopes of human neural excitation patterns may be estimated via the second cochlea map and human masking data (Wegel and Lane, 1924). In the present study, slopes are estimated from limited existing psychophysical data, which can be improved with more accurate experimental data in the future.

  PublisherDOI

• On causality and aural impulse responses synthesized using the inverse discrete Fourier transform

  The Journal of the Acoustical Society of America · 2021 · 5 citations

  • Computer Science
  • Mathematics
  • Computer Science

  Causality is a fundamental property of physical systems and dictates that a time impulse response characterizing any causal system must be one-sided. However, when synthesized using the inverse discrete Fourier transform (IDFT) of a corresponding band-limited numerical frequency transfer function, several papers have reported two-sided IDFT impulse responses of ear-canal reflectance and ear-probe source parameters. Judging from the literature on ear-canal reflectance, the significance and source of these seemingly non-physical negative-time components appear largely unclear. This paper summarizes and clarifies different sources of negative-time components through ideal and practical examples and illustrates the implications of constraining aural IDFT impulse responses to be one-sided. Two-sided IDFT impulse responses, derived from frequency-domain measurements of physical systems, normally occur due to the two-sided properties of the discrete Fourier transform. Still, reflectance IDFT impulse responses may serve a number of practical and diagnostic purposes.

  PublisherDOI

• Stream 1: Number Systems

  2020-01-01

  book-chapter1st authorCorresponding
  PublisherDOI

• Stream 3B: Vector Calculus

  2020-01-01

  book-chapter1st authorCorresponding
  PublisherDOI

• Understanding Interdependencies between Mechanical Velocity and Electrical Voltage in Electromagnetic Micromixers

  Micromachines · 2020 · 12 citations

  Senior authorCorresponding
  • Mechanical engineering
  • Electrical engineering
  • Electronic engineering

  Micromixers are critical components in the lab-on-a-chip or micro total analysis systems technology found in micro-electro-mechanical systems. In general, the mixing performance of the micromixers is determined by characterising the mixing time of a system, for example the time or number of circulations and vibrations guided by tracers (i.e., fluorescent dyes). Our previous study showed that the mixing performance could be detected solely from the electrical measurement. In this paper, we employ electromagnetic micromixers to investigate the correlation between electrical and mechanical behaviours in the mixer system. This work contemplates the "anti-reciprocity" concept by providing a theoretical insight into the measurement of the mixer system; the work explains the data interdependence between the electrical point impedance (voltage per unit current) and the mechanical velocity. This study puts the electromagnetic micromixer theory on a firm theoretical and empirical basis.

  PublisherOA PDFDOI

• Stream 3A: Scalar Calculus

  2020-01-01

  book-chapter1st authorCorresponding
  PublisherDOI

• An Invitation to Mathematical Physics and Its History

  Springer eBooks · 2020 · 7 citations

  1st authorCorresponding
  • Computer Science
  • Theoretical physics
  • Computer Science
  PublisherDOI

Recent grants

• NIH Grant R21DC009277

  NIH · $393k · 2011

Frequent coauthors

• Patricia S. Jeng

  University of Illinois Urbana-Champaign

  17 shared

• Feipeng Li

  Xi'an Jiaotong University

  14 shared

• P. F. Fahey

  University of Scranton

  13 shared

• Stephen T. Neely

  Carnegie Mellon University

  13 shared

• Woojae Han

  Hallym University

  11 shared

• Sunil Puria

  Harvard University

  10 shared

• L. R. Rabiner

  8 shared

• Robert H. Withnell

  Indiana University Bloomington

  8 shared

Awards & honors

• Alumni Award for Distinguished Service