A simulation framework for auditory discrimination experiments: Revealing the importance of across-frequency processing in speech perception

Abstract

A framework for simulating auditory discrimination experiments, based on an approach from Schädler, Warzybok, Hochmuth, and Kollmeier [(2015). Int. J. Audiol. 54, 100-107] which was originally designed to predict speech recognition thresholds, is extended to also predict psychoacoustic thresholds. The proposed framework is used to assess the suitability of different auditory-inspired feature sets for a range of auditory discrimination experiments that included psychoacoustic as well as speech recognition experiments in noise. The considered experiments were 2 kHz tone-in-broadband-noise simultaneous masking depending on the tone length, spectral masking with simultaneously presented tone signals and narrow-band noise maskers, and German Matrix sentence test reception threshold in stationary and modulated noise. The employed feature sets included spectro-temporal Gabor filter bank features, Mel-frequency cepstral coefficients, logarithmically scaled Mel-spectrograms, and the internal representation of the Perception Model from Dau, Kollmeier, and Kohlrausch [(1997). J. Acoust. Soc. Am. 102(5), 2892-2905]. The proposed framework was successfully employed to simulate all experiments with a common parameter set and obtain objective thresholds with less assumptions compared to traditional modeling approaches. Depending on the feature set, the simulated reference-free thresholds were found to agree with-and hence to predict-empirical data from the literature. Across-frequency processing was found to be crucial to accurately model the lower speech reception threshold in modulated noise conditions than in stationary noise conditions.

Influence of auditory attention on sentence recognition captured by the neural phase

“…These models can predict speech intelligibility in different listening situations based on individual auditory profiles and on various auditory features (e.g. Rhebergen et al ., ; Meyer & Brand, ; Schädler et al ., ). However, they do not consider cognitive abilities, such as selective auditory attention, as a parameter in estimating speech recognition.…”

Microscopic Multilingual Matrix Test Predictions Using an ASR-Based Speech Recognition Model

“…In the current study, a framework for auditory discrimination experiments [2,8] which uses an automatic speech recognition (ASR) system and that neither requires calibration with empirical data nor the temporal alignment of the to-be recognized signal was used as a microscopic model to simulate and hence predict the outcome of the matrix test across several languages and noise conditions as empirically determined in [6]. FADE was shown to accurately predict speech intelligibility of the German matrix test in different stationary noise conditions [7].…”

“…FADE was shown to accurately predict speech intelligibility of the German matrix test in different stationary noise conditions [7]. Its scope was then successfully extended to a fluctuating noise conditions and even basic psychoacoustical experiments [8]. The matrix sentence test was developed for several languages in order to make speech recognition measurements as comparable across languages as possible.…”

Modelling human speech recognition in challenging noise maskers using machine learning

“…In comparison with the recently published FADE model [20] which uses a GMM/HMM recognizer specifically trained on the same Matrix test material as used for the prediction, the current DNN/HMM model employs a more sophisticated recognizer that has not seen the exact speech token under test before and could potentially operate on an open speech recognition set as well. However, this comes at the cost of more training data needed for this approach and further restrictions like much more computational resources required of the model parameters for the respective experiment to be modelled.…”

“…However, this comes at the cost of more training data needed for this approach and further restrictions like much more computational resources required of the model parameters for the respective experiment to be modelled. In addition, the current model approach has not been tested yet for predicting psychoacoustic experiments and the effect of hearing impairment in a similar way as successfully implemented with FADE [20].…”