Information Loss in the Human Auditory System

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

From the eardrum to the auditory cortex, where acoustic stimuli are decoded, there are several stages of auditory processing and transmission where information may potentially be lost. In this paper, we aim at quantifying the total information loss in the human auditory system by using information theoretic tools. To do so, we consider a speech communication model, where words are uttered and sent through a noisy channel, and then received and processed by a human listener. We define a notion of information loss that is related to the human word recognition rate. To assess the word recognition rate of humans, we conduct a closed-vocabulary intelligibility test. We derive upper and lower bounds on the information loss. Simulations reveal that the bounds are tight and we observe that the information loss in the human auditory system increases as the signal to noise ratio (SNR) decreases. Our framework also allows us to study whether humans are optimal in terms of speech perception in a noisy environment. Toward that end, we derive optimal classifiers and compare the human and machine performance in terms of information loss and word recognition rate. We observe a higher information loss and lower word recognition rate for humans compared to the optimal classifiers. In fact, depending on the SNR, the machine classifier may outperform humans by as much as 8 dB. This implies that for the speech-in-stationary-noise setup considered here, the human auditory system is suboptimal for recognizing noisy words.

Original languageEnglish
Article number8579632
JournalIEEE/ACM Transactions on Audio, Speech, and Language Processing
Volume27
Issue number3
Pages (from-to)472-481
Number of pages10
ISSN2329-9290
DOIs
Publication statusPublished - Mar 2019

Fingerprint

classifiers
Classifiers
Signal to noise ratio
signal to noise ratios
eardrums
Speech communication
human performance
intelligibility
cortexes
data transmission
stimuli
Acoustics
communication
Processing
acoustics
simulation

Keywords

  • Gaussian mixture model
  • Human auditory system
  • maximum likelihood classifier
  • mutual information

Cite this

@article{28c24478120e43829ecc4972df890dfb,
title = "Information Loss in the Human Auditory System",
abstract = "From the eardrum to the auditory cortex, where acoustic stimuli are decoded, there are several stages of auditory processing and transmission where information may potentially be lost. In this paper, we aim at quantifying the total information loss in the human auditory system by using information theoretic tools. To do so, we consider a speech communication model, where words are uttered and sent through a noisy channel, and then received and processed by a human listener. We define a notion of information loss that is related to the human word recognition rate. To assess the word recognition rate of humans, we conduct a closed-vocabulary intelligibility test. We derive upper and lower bounds on the information loss. Simulations reveal that the bounds are tight and we observe that the information loss in the human auditory system increases as the signal to noise ratio (SNR) decreases. Our framework also allows us to study whether humans are optimal in terms of speech perception in a noisy environment. Toward that end, we derive optimal classifiers and compare the human and machine performance in terms of information loss and word recognition rate. We observe a higher information loss and lower word recognition rate for humans compared to the optimal classifiers. In fact, depending on the SNR, the machine classifier may outperform humans by as much as 8 dB. This implies that for the speech-in-stationary-noise setup considered here, the human auditory system is suboptimal for recognizing noisy words.",
keywords = "Gaussian mixture model, Human auditory system, maximum likelihood classifier, mutual information",
author = "Jahromi, {Mohsen Zareian} and Adel Zahedi and Jesper Jensen and Jan {\O}stergaard",
year = "2019",
month = "3",
doi = "10.1109/TASLP.2018.2882913",
language = "English",
volume = "27",
pages = "472--481",
journal = "IEEE/ACM Transactions on Audio, Speech, and Language Processing",
issn = "2329-9290",
publisher = "IEEE Signal Processing Society",
number = "3",

}

Information Loss in the Human Auditory System. / Jahromi, Mohsen Zareian; Zahedi, Adel; Jensen, Jesper; Østergaard, Jan.

In: IEEE/ACM Transactions on Audio, Speech, and Language Processing, Vol. 27, No. 3, 8579632, 03.2019, p. 472-481.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Information Loss in the Human Auditory System

AU - Jahromi, Mohsen Zareian

AU - Zahedi, Adel

AU - Jensen, Jesper

AU - Østergaard, Jan

PY - 2019/3

Y1 - 2019/3

N2 - From the eardrum to the auditory cortex, where acoustic stimuli are decoded, there are several stages of auditory processing and transmission where information may potentially be lost. In this paper, we aim at quantifying the total information loss in the human auditory system by using information theoretic tools. To do so, we consider a speech communication model, where words are uttered and sent through a noisy channel, and then received and processed by a human listener. We define a notion of information loss that is related to the human word recognition rate. To assess the word recognition rate of humans, we conduct a closed-vocabulary intelligibility test. We derive upper and lower bounds on the information loss. Simulations reveal that the bounds are tight and we observe that the information loss in the human auditory system increases as the signal to noise ratio (SNR) decreases. Our framework also allows us to study whether humans are optimal in terms of speech perception in a noisy environment. Toward that end, we derive optimal classifiers and compare the human and machine performance in terms of information loss and word recognition rate. We observe a higher information loss and lower word recognition rate for humans compared to the optimal classifiers. In fact, depending on the SNR, the machine classifier may outperform humans by as much as 8 dB. This implies that for the speech-in-stationary-noise setup considered here, the human auditory system is suboptimal for recognizing noisy words.

AB - From the eardrum to the auditory cortex, where acoustic stimuli are decoded, there are several stages of auditory processing and transmission where information may potentially be lost. In this paper, we aim at quantifying the total information loss in the human auditory system by using information theoretic tools. To do so, we consider a speech communication model, where words are uttered and sent through a noisy channel, and then received and processed by a human listener. We define a notion of information loss that is related to the human word recognition rate. To assess the word recognition rate of humans, we conduct a closed-vocabulary intelligibility test. We derive upper and lower bounds on the information loss. Simulations reveal that the bounds are tight and we observe that the information loss in the human auditory system increases as the signal to noise ratio (SNR) decreases. Our framework also allows us to study whether humans are optimal in terms of speech perception in a noisy environment. Toward that end, we derive optimal classifiers and compare the human and machine performance in terms of information loss and word recognition rate. We observe a higher information loss and lower word recognition rate for humans compared to the optimal classifiers. In fact, depending on the SNR, the machine classifier may outperform humans by as much as 8 dB. This implies that for the speech-in-stationary-noise setup considered here, the human auditory system is suboptimal for recognizing noisy words.

KW - Gaussian mixture model

KW - Human auditory system

KW - maximum likelihood classifier

KW - mutual information

UR - http://www.scopus.com/inward/record.url?scp=85059371570&partnerID=8YFLogxK

U2 - 10.1109/TASLP.2018.2882913

DO - 10.1109/TASLP.2018.2882913

M3 - Journal article

VL - 27

SP - 472

EP - 481

JO - IEEE/ACM Transactions on Audio, Speech, and Language Processing

JF - IEEE/ACM Transactions on Audio, Speech, and Language Processing

SN - 2329-9290

IS - 3

M1 - 8579632

ER -