Amplitude modulations in the speech convey important acoustic information for speech perception. Auditory steady state response (ASSR) is thought to be physiological correlate of amplitude modulation perception. Limited research is available exploring association between ASSR and modulation detection ability as well as speech perception. Correlation of modulation detection thresholds (MDT) and speech perception in noise with ASSR was investigated in twofold experiments. 30 normal hearing individuals and 11 normal hearing individuals within age range of 18–24 years participated in experiments 1 and 2, respectively. MDTs were measured using ASSR and behavioral method at 60 Hz, 80 Hz, and 120 Hz modulation frequencies in the first experiment. ASSR threshold was obtained by estimating the minimum modulation depth required to elicit ASSR (ASSR-MDT). There was a positive correlation between behavioral MDT and ASSR-MDT at all modulation frequencies. In the second experiment, ASSR for amplitude modulation (AM) sweeps at four different frequency ranges (30–40 Hz, 40–50 Hz, 50–60 Hz, and 60–70 Hz) was recorded. Speech recognition threshold in noise (SRTn) was estimated using staircase procedure. There was a positive correlation between amplitude of ASSR for AM sweep with frequency range of 30–40 Hz and SRTn. Results of the current study suggest that ASSR provides substantial information about temporal modulation and speech perception.
Speech acoustics have multiple temporal characteristics [
Temporal envelope of speech can be considered as a complex amplitude modulation, which is a sum of many modulators. Modulation filter banks located in the auditory system split the complex modulations into series of sinusoidal modulations [
All these lines of evidence suggest that TMTF provides valuable information related to speech perception. However, TMTF has to be measured using behavioral paradigms in which the active cooperation of the subject is required. Therefore it becomes challenging while testing the “difficult to test population”. For this reason, there is a need for objective tool for obtaining MDTs. Purcell et al. [
The sweep techniques used by Purcell et al. [
A total of 30 normal hearing individuals (25 females, 5 males) within age range of 18–24 years (mean age = 21 years) participated in experiment 1. 11 normal hearing individuals within age range of 18–24 years participated in experiment 2. All participants were selected using nonrandom sampling technique. The subjects included for the study had audiograms demonstrative of normal hearing thresholds (<15 dBHL pure tone thresholds for octave frequencies from 0.25 to 8 kHz). The participants had a normal middle ear functioning, with “A” type tympanogram and ipsi- and contralateral stapedial reflexes present at 500, 1000, 2000, and 4000 Hz. Subjects with a history of otologic or neurologic diseases or with auditory processing deficits were excluded from the study. All the participants were recruited with an informed consent prior to the conduction of the study. The study protocol was approved by the institutional ethical committee. Data was collected at Department of Audiology, Kasturba Medical College, Mangalore, over duration of March 2012 to February 2013.
For recording and analyzing ASSR, IHS SmartEP ASSR version 3.92 was used. MATLAB version 7.0 was used to generate and present the signal/stimulus for behavioral estimation of modulation detection thresholds which was routed to GSI-61 clinical audiometer.
Broad band white noise was created with a sampling rate of 20000 Hz, which was then filtered between 100 and 7999 Hz using 4th butterworth filter. Broadband noise carrier was refreshed on each presentation. Total duration of the stimulus was one second. Sinusoidal modulators with 60 Hz, 80 Hz, and 120 Hz frequencies were then created with a starting phase of zero degree. Relatively high rates of amplitude modulations were used in the current study and modulation rates in these are necessary for stream segregation [
Sinusoidal sweeping chirps were created with a sampling frequency of 20,000 Hz. These sweeping chirps stimuli were used to amplitude modulate a band limited white noise with the bandwidth of 100–7999 Hz. Stimuli with sweeping amplitude modulations were created for four different frequency ranges including 30–40 Hz, 40–50 Hz, 50–60 Hz, and 60–70 Hz. Stimuli had a total duration of 1 sec which comprised 100 msec unmodulated segment at initial and final position. Middle 800 ms segment was modulated.
Tenlists of HINT [
The white noise which is amplitude modulated at 60 Hz, 80 Hz, and 120 Hz was used as stimulus. The stimuli were presented using customized program written in MATLAB which were routed through GSI-61 clinical audiometer. Stimuli were presented at 70 dBSPL to the right ear through TDH 39 headphones. Experiments were performed in sound treated audiometric room. Two-down one-up procedure [
Intelligent hearing system (IHS) version 3.92 Smart ASSR was used to record the evoked responses. The subject was seated on a comfortable reclining chair in a sound treated room and was asked to be relaxed throughout the recording session in order to minimize the artifacts. The electrode sites were cleaned using a skin prepping gel and AgCl electrodes were placed using the conventional single channel montage with inverting electrode placed on ipsilateral (right) mastoid, noninverting to vertex and ground on the contralateral (left) mastoid. Absolute electrode impedance and intraelectrode impedance were less than 5000 Ohms and 2000 Ohms, respectively. The white noise which is amplitude modulated at 60 Hz, 80 Hz, and 120 Hz was presented at 70 dBSPL in the right ear through Etymotic ER-3A insert earphones. Responses were elicited at different modulation depths at each modulation frequency. The response is determined automatically by the instrument using frequency weighted averaging method, where “
The procedure was similar to estimation of ASSR-MDT. However, responses were estimated at fixed modulation depth of 100%. Presentation level was 70 dB SPL. Responses for stimuli with sweeping amplitude modulations at four different frequency ranges including 30–40 Hz, 40–50 Hz, 50–60 Hz, and 60–70 Hz were elicited for the right ear. A total of 200 sweeps were recorded for each stimulus. Each sweep lasted for 1 second. Two recordings were taken at each modulation frequency range.
The subject’s speech recognition threshold in noise (SRTn) was obtained by adjusting the speech-to-noise ratio (SNR). This was achieved by keeping the speech level constant and by reducing the root mean square level of noise. SNR was varied in 2 dB steps using staircase procedure [
ASSR-MDT was determined by obtaining the minimum modulation depth at which the ASSR could be recorded. This was performed at three different modulation frequencies (i.e., at 60 Hz, 80 Hz, and 120 Hz). The MDTs were determined behaviorally for each subject at these modulation frequencies using transformed up-down procedure. The results from electrophysiological method (ASSR) were compared to behavioral measures of the TMTF. The mean MDTs for 60 Hz, 80 Hz, and 120 Hz modulation frequencies obtained using ASSR and behavioral measures are given in Table
Mean and standard deviation of ASSR and behavioral MDT.
Modulation frequency |
ASSR MDT (%) | Behavioral MDT (%) | ||
---|---|---|---|---|
Mean | SD | Mean | SD | |
60 | 19.5 | 9.32 | 23.35 | 7.28 |
80 | 22.33 | 9.17 | 27.81 | 6.42 |
120 | 26.67 | 10.77 | 34.98 | 10.68 |
Both behavioral and ASSR MDTs are expressed in percentages.
Mean and standard deviation of ASSR amplitude in dB (
Modulation frequency |
Mean (dB) | Standard deviation |
---|---|---|
60 | 19.06 | 7.76 |
80 | 18.3 | 9.75 |
120 | 17.49 | 7.71 |
It can be seen from Table
TMTF constructed using mean MDT (dB) obtained using ASSR (rectangles) and behavioral method (circles). (MDT (dB) =
Pearson’s correlation analysis was used to investigate the association between behavioral and ASSR modulation detection thresholds. The results revealed a significant positive correlation between ASSR modulation detection threshold and behavioral thresholds at 60 Hz (
Scatterplot representing association between ASSR MDT and behavioral MDT at 60 Hz, 80 Hz, and 120 Hz modulation frequencies.
Predictability of behavioral MDT using ASSR MDT was assessed using linear regression analysis. It was found that for 60 Hz modulation frequency about 60% of variance in behavioral threshold can be attributed to variance observed in ASSR thresholds (
Regression equations to obtain behavioral MDT from ASSR MDT at each modulation frequency.
60 Hz | 80 Hz | 120 Hz |
---|---|---|
|
|
|
Grand average response was derived for each stimulus by summing the response of all subjects. The grand averaged response was subjected to the time frequency analysis. Short time Fourier transform (STFT) was done to analyze the responses in time frequency domain. Analysis was done at 1024-point frequency bin and a hamming window was used to smooth the frequency response. Results of the STFT were represented graphically. STFT analysis confirmed the coding of modulation sweeps at auditory system which provides a physiological evidence for envelope tracking ability of auditory system. Results of the STFT analysis are presented in the form of spectrograms in Figures
Short time Fourier analysis of the grand averaged response for the frequency range of 30–40 Hz.
Short time Fourier analysis for the grand averaged response for the frequency range of 40–50 Hz.
Short time Fourier analysis for the grand averaged response for the frequency range of 50–60 Hz.
Short time Fourier analysis for the grand averaged response for the frequency range of 60–70 Hz.
Power analysis in the frequency range of amplitude modulation sweeps was performed for the recorded responses. FFT analysis was performed to identify the evoked responses in the frequency region of the AM sweeps and their corresponding amplitude. RMS amplitudes of the evoked responses in the frequency regions were then calculated for each modulation sweep range. These RMS amplitudes were subjected to further statistical analysis. Mean and standard deviation values for the amplitude of the evoked responses are given in Table
Mean and standard deviation of ASSR amplitude (
Stimulus | Mean RMS amplitude ( |
SD ( |
---|---|---|
30–40 Hz sweep | 3.90 | 0.96 |
40–50 Hz sweep | 3.57 | 0.99 |
50–60 Hz sweep | 3.05 | 1.31 |
60–70 Hz sweep | 4.09 | 2.24 |
Amplitude of ASSR for AM sweeps was correlated with SRTn to investigate the relationship between speech recognition ability and ASSR. Pearson’s correlation analysis was used to assess the possible association between ASSR for AM sweeps and SRTn. The results revealed a significant positive correlation between ASSR for AM sweep and SRTn only at 30–40 Hz range (
Based on modulation detection thresholds, the psychophysical temporal modulation rate transfer function (MTF) exhibits a low pass characteristic with MDTs declining with increasing modulation rate. Normal hearing listeners have low threshold for slow modulations and threshold increases as the modulation rate is increased in the TMTF task [
Moderate-to-strong positive correlation was observed between the ASSR and behavioral modulation detection thresholds. Also there was a linear relationship between ASSR-MDT and behavioral MDT. ASSR can be considered as physiological equivalent of subjective modulation perception. Hence, it is reasonable to expect a correlation between these two. Association between ASSR-MDT and behavioral MDT can be explained by relating the ASSR generation to model for amplitude modulation perception (Figure
Schematic diagram explaining the mechanism of temporal modulation perception. Mechanism explained here is based on modulation filter bank (MFB) model proposed by Dau et al. [
Mean ASSR thresholds obtained in the current study were slightly smaller than behavioral thresholds. Response bias in the decision making process may be the possible reason for this observed difference. Similarly, trends have been shown in previous attempts to objectively assess the temporal acuity. Werner et al. [
STFT analysis confirmed the ability of the auditory system to code sweeping modulation, which provides the physiological evidence for envelope tracking ability of auditory system. There was positive correlation observed between strength of AM sweep coding and speech perception in noise. Response latency, precision of response timing, and response magnitude of specialized IC neurons are the important factors for tracking envelope changes over time [
The current study evaluated the utility of ASSR as an objective tool for assessment of temporal modulation perception and speech perception. The first experiment investigated the association between MDT measured using ASSR and behavioral method at 60 Hz, 80 Hz, and 120 Hz. The results of this experiment indicated that there are a strong correlation at 60 Hz and moderate correlation at 80 Hz and at 120 Hz. This suggests that the MDT using ASSR could serve as an objective measure of temporal resolution, which is well correlated with the behavioral measurements. The second experiment explored the association between envelope following response (ASSR) for amplitude modulation sweeps and speech perception in noise. Short time Fourier transform (STFT) analysis confirmed the ability of the auditory system to code sweeping modulation, which provides the physiological evidence for envelope tracking ability of auditory system. The results from ASSR using AM sweep and speech recognition threshold in noise (SRTn) showed a positive correlation between strength of AM sweep coding in 30–40 Hz range and speech perception.
The authors report no potential conflict of interests involving financial interests. Part of this research is submitted to the University of Manipal as part of postgraduate master’s dissertation.