**Glossary**

#### **Human hearing research**

*Auditory masking*: perceptual phenomenon that occurs when the threshold of audibility for one sound is raised in the presence of another sound.

*Cochlear implant*: a surgically implanted electronic device that restores a sense of hearing.

*Critical band theory*: estimates the bandwidth of spectral frequencies within which a second sound is predicted to interfere with the perception of the first sound by auditory masking.

*Electric*‐*on*‐*acoustic masking*: reproduction of masking by using cochlear implant electrodes.

*Electric*‐*on*‐*electric masking*: production of masking using only cochlear implant electrodes.

*Energy detector model*: the nonlinear power spectrum model approximation of auditory responses (which is the inspiration for acoustic features such as FBANK and MFCC).

*Place theory*: pitch perception depends on the location along the basilar membrane.

*Profile analysis*: a signal is detected by noting a change in the spectrum at some frequency.

*Temporal critical bands*: critical bandwidth for a temporal process (e.g., temporal envelope).

*Temporal theory*: pitch perception depends on the temporal firing patterns of neurons.

*Transition bandwidths*: occurrence of an interplay between spectral and temporal processes.

*Volley theory*: groups of neurons respond to a sound by firing action potentials slightly out‐ of-phase to encode a greater representation of sound that is sent to the brain.

#### **Machine hearing research**

*Automatic speech recognition*: a computational method that allows recognition of language. *Data compression*: algorithm that reduces the audio transmission and storage requirements. *Deep learning*: branch of machine learning that models high level abstractions in the data via multiple layers of processing and nonlinear transformations within hierarchal structures. *Deep neural networks*: artificial neural network inspired by the hierarchal modeling of brains. *Double*‐*BANK (*⧠⧠*) features*: the combination of features (e.g.: FBANK+TBANK) as inspired by the observance of temporal critical bands and transition bandwidths in the auditory system. *Frequency amplitude modulation encoding (FAME)*: alternative features derived via TBANK. *Gaussian mixture model hidden Markov model (GMM*‐*HMM)*: Bayesian method to align DNNs. *Spectral filterbanks*: acoustic features (MFCC or FBANK) inspired by the nonlinear spacing of power spectrum or energy detector models, and critical band theory. *Temporal filterbanks*: acoustic features (e.g.: TBANK) inspired by temporal critical bands.
