The spectro-temporal ripple for investigating processor effectiveness (STRIPES) test is a psychophysical measure of spectro-temporal resolution in cochlear-implant (CI) listeners. It has been validated using direct-line input and loudspeaker presentation with listeners of the Advanced Bionics CI. This article investigates the suitability of an online application using wireless streaming (webSTRIPES) as a remote test. It reports a strong across-listener correlation between STRIPES thresholds obtained using laboratory testing with loudspeaker presentation vs remote testing with streaming presentation, with no significant difference in STRIPES thresholds between the two measures. WebSTRIPES also produced comparable and robust thresholds with users of the Cochlear CI.

The widespread social distancing necessitated by the COVID-19 pandemic has led to a greater need for remote, online testing platforms in many scientific disciplines, including cochlear-implant (CI) listener behavioral psychophysics. Successful implementation of an online psychophysical test for CI users would present many new opportunities for research and would increase the geographical area from which participants can be recruited, as they would no longer be required to travel to a laboratory to participate in a research study (De Graaff et al., 2018; Peng et al., 2020). Such a test might also be implemented clinically to monitor changes in auditory performance without requiring the patient to attend in person.

An obstacle encountered by online tests of acoustic hearing concerns the widely varying quality of headphones and sound cards used by participants (Goehring et al., 2012). Fortunately, the ability of modern CIs to stream audio directly from internet-enabled devices (including smartphones and tablets), combined with the homogeneity of each manufacturer's device, largely overcomes this problem. As a result, stimulus presentation in online tests can be more tightly controlled than in equivalent normal-hearing (NH) or hearing-impaired (HI) listener studies.

Speech intelligibility performance can vary greatly among CI listeners, especially in noisy situations. To address this, many new methods for improving CI listener performance, including new stimulation modes, individualized fitting techniques, and novel processing strategies, have been developed (see Carlyon and Goehring, 2021). As speech tests rely heavily on listener acclimatization over timescales much longer than one clinical appointment (Davis et al., 2005; Davis and Johnsrude, 2007), new tests to rapidly and accurately assess novel methods have been sought. Several recent studies have discussed the potential limitation that the listener's clinical strategy has a systematic advantage over novel strategies, due to long-term acclimatization to the former (Croghan et al., 2017; Schvartz-Leyzac et al., 2017; Berg et al., 2019). At worst, this can lead to false conclusions about the performance of potentially beneficial new strategies or methods and prevent further investigation of them. Take-home evaluation of a new method partially solves this issue but may be unethical if the new program is unsuccessful, as it exposes the listener to long periods of poor hearing. To negate the need for long-term acclimatization and take-home evaluation, several non-speech tests have been developed to obtain quick, reliable information on improvements to listener auditory perception due to a novel CI method (Supin et al., 1994; Henry and Turner, 2003; Litvak et al., 2007; Won et al., 2007; Drennan et al., 2008; Saoji et al., 2009; Won et al., 2011; Azadpour and McKay, 2012; Aronoff and Landsberger, 2013). One such test is STRIPES (Spectro-Temporal Ripple for Investigating Processor Effectiveness), which has been developed via several iterations for CIs manufactured by Advanced Bionics (AB) (Los Angeles, CA) (Archer-Boyd et al., 2018; Archer-Boyd et al., 2020). The test presents listeners with concurrent sinusoidal sweeps that either increase or decrease in frequency over time and requires listeners to discriminate between downward- and upward-sweeping stimuli (Fig. 1). Increasing the temporal overlap between each sweep increases the difficulty. Amplitude modulation (AM) at any one frequency is identical between upward and downward sweeps at the same density. It was designed to meet a number of criteria, as described in Archer-Boyd et al. (2018). These include simultaneous spectral and temporal processing being needed for good performance and good discrimination being impossible using only one local spectro-temporal segment of each stimulus. The test should instead require extraction of higher-order cues from the stimuli, in this case the direction of frequency sweeps. Initially, the STRIPES stimuli were designed to remove any within-channel cues other than spectro-temporal modulation that were visible in the electrodogram and could potentially be used to perform the task. The method of presentation (via a direct audio input cable) was tightly controlled. A later study (Archer-Boyd et al., 2020) relaxed the design requirements and controls on stimulus, CI processing, and presentation method and found that listeners' thresholds were not significantly different across conditions compared to the direct-stimulation method. Test results were also shown to be robust over time, with no significant difference between thresholds obtained on different days. STRIPES successfully predicted the best speech-in-noise performance at the level of individual listeners of two experimental fitting algorithms based on deactivating subsets of electrodes (Goehring et al., 2019). It has also been shown to correlate with individual measures of channel interaction using spectrally blurred speech-in-noise stimuli (Goehring et al., 2020).

Fig. 1.

Schematic representation of the spectrograms of the three stimuli present in one trial of the STRIPES test. In this example, the signal, consisting of upward sweeps, is present in interval 1, and the stripe density is 5.

Fig. 1.

Schematic representation of the spectrograms of the three stimuli present in one trial of the STRIPES test. In this example, the signal, consisting of upward sweeps, is present in interval 1, and the stripe density is 5.

Close modal

The present study compares performance on different implementations of the STRIPES test. One of these is a loudspeaker-presented, lab-based version (“loudspeaker STRIPES”) using data presented by Archer-Boyd et al. (2020). The second is a new online implementation (“webSTRIPES”) that uses streaming via a wireless connection between a participant's phone/tablet/laptop and their CI speech processor. The data from these conditions were obtained from the same users of the AB CI device. Finally, users of Cochlear (Sydney, Australia) CI devices were tested using the webSTRIPES method to determine whether robust and reliable thresholds could be obtained, as previously observed for the AB CI device.

The online implementation of the STRIPES test procedure is very similar to the implementation by Archer-Boyd et al. (2020) (“loudspeaker STRIPES”) and is described briefly here, highlighting the changes made. The test measures the spectral-density threshold at which the target stimulus can be just distinguished from two reference stimuli by the listener in a three-interval, two-alternative forced-choice task in an adaptive staircase procedure. Stimuli are created from 1-s-long, concurrent and concatenated exponential sine sweeps (with random starting phase) moving up or down in frequency from 250 to 8000 Hz, at a rate of 5 octaves/s. Sweep density is defined as the number of concurrent sweeps present at one time. Non-integer density values are possible; for example, a density of 2.5 would mean that for 50% of the time, two swept sinusoids were present simultaneously (overlapped) and that for the other 50% of the time, three swept sinusoids overlapped. The delay between the start of each sweep is 1/density, such that at density = 5, the delay between the start of each sweep is 200 ms. As the bandwidth of the stimulus is 5 octaves, the number of ripples per octave (RPO) is equal to density/5; the spectrograms shown in Fig. 1 are for a density of 5 and RPO = 1. As the individual sweeps are 1 s in duration, the AM frequency in Hz is the same as the density. An equation for a single cycle of a STRIPES stimulus is given in Archer-Boyd et al. (2020). The number of cycles, and hence the duration of the stimuli, decreases with increasing density, such that at least two unbroken single sweeps are presented in each stimulus regardless of the starting phase and that an integer number of cycles at each density is always presented. The duration of the STRIPES stimuli varied from 2.07 s (density = 1.1) to approximately 1.3 s at the highest densities presented. Ensuring an integer number of cycles also has the effect of matching the start and end instantaneous frequencies for each stimulus, minimizing this potentially confusing and salient stimulus feature.

The listener selects the target interval, which is either the first or last interval and is always an upward sweep; the other two intervals contain downward sweeps. The starting phases of the stimuli are randomized. The sweep density is varied to titrate difficulty, with the task being easy at a density close to 1 peak per 5-octave sweep and progressively harder at higher densities. The starting frequency is roved across trials, and the beginning and end of each interval are masked by short noise bursts to reduce the salience of onset and offset cues. To simplify the test delivery, the webSTRIPES stimuli, although initially identical to those used in Archer-Boyd et al. (2020), are drawn from a library of pre-made stimuli with ten fixed starting-frequency rove values spread equally over one cycle, rather than the continuous rove used previously. The noise bursts are also included in the pre-made stimuli, rather than being independently generated for each trial. Finally, to reduce the volume of data downloaded by the listener's device, reducing the potential for failed tests due to slow or intermittent internet, the stimuli are 96 kbps mp3 files instead of lossless WAV files. This produces little to no audible effect on the stimuli or visible artefacts in their spectrograms. An adaptive two-up/one-down procedure starts with a sweep density of 1.1 (number of sweeps concurrently presented during each sweep) and adjusts the density per trial with a density step size of 0.5 (for the first four reversals) and 0.2 (for the last four reversals). The inter-stimulus interval was changed to 1 s, instead of the 0.6 s used previously, due to playback timing issues in some web browsers. The test ends after eight reversals (instead of the previous 12) to reduce the length of each test by 26.8%, and the final threshold of the run was calculated as the average of the last four reversals.

Previous versions of the test used a custom graphical user interface (GUI) and code written in matlab (MathWorks, 2019) for presentation via either a direct audio connection from a laptop and soundcard to a CI speech processor or via a loudspeaker in a soundproof booth. In contrast, webSTRIPES was implemented in Just Another Tool For Online Studies (JATOS), an open-source, cross-platform web application for hosting online studies written in javascript (Lange et al., 2015). The test was hosted on a local server at the University of Cambridge.

The signal paths of loudspeaker STRIPES and webSTRIPES differed in several ways. The loudspeaker version presented uncompressed WAV files via matlab and a lab laptop, through an RME (Haimhausen, Germany) UCX soundcard and Genelec (Iisalmi, Finland) 8020c loudspeaker. Tests were conducted in a soundproof booth with the participant seated approximately 1 m from the loudspeaker and free to move their head relative to the loudspeaker. The webSTRIPES version used compressed mp3 files presented via a web browser and a lab laptop or the participant's laptop, tablet, or phone, via Bluetooth to a proprietary Bluetooth receiver (the exact model differing across participants). Tests were conducted in the participants' home environments or in the lab office space, and the mixing ratio between the Bluetooth and microphone signals could be 50%/50% or 70%/30%, depending on how the processor was initially set up by an audiologist.

Nine of ten AB users tested previously in Archer-Boyd et al. (2020) took part. All used a Naida CI Q90 speech processor implementing the HiRes Optima-S strategy, which was the same as used in the clinical Optima S condition of Archer-Boyd et al. (2020), where further participant and processor details can be found. One participant was unavailable for testing. Six Cochlear Corp. users also took part and are described in Table 1.

Table 1.

Information on participants who used the Cochlear device. All used the ACE processing strategy. For information on the AB users, the reader is referred to Archer-Boyd et al. (2020).

ParticipantAgeImplant
C09 70 CI24RE (CA) 
C26 58 CI522 
C27 71 CI512 
C29 73 CI522 
C30 73 CI512 
C34 69 CI24RE 
ParticipantAgeImplant
C09 70 CI24RE (CA) 
C26 58 CI522 
C27 71 CI512 
C29 73 CI522 
C30 73 CI512 
C34 69 CI24RE 

Participants accessed the test via their chosen web browser using any suitable Internet-enabled device (e.g., smartphone, laptop, tablet) of their choice. They were sent a unique participant uniform resource locator (url) that could be used multiple times. An example test can be run by pasting the following url into a web browser: https://lsr-studies-02.mrc-cbu.cam.ac.uk/publix/zgUz5wMrg2a (note that the task should initially be very easy for a NH person listening acoustically). At the beginning of the test, participants were instructed to connect their internet-enabled device wirelessly to their CI processor as they would normally and set the volume to a comfortable level, using a repeating density = 5 STRIPES stimulus presented via the application. After this, the online experiment proceeded in a similar way to previous lab-based versions. For training, participants were presented with six trials at the lowest density (1.1), and the correct answer was highlighted in green. After this, the correct answer was not highlighted, and listeners were presented with five trials at the same density (pre-test screening). A score of 4 or 5 correct led to the main (adaptive staircase) experiment. A score of <4 resulted in presentation of the pre-test screening again until the participant scored 4 or more correct. If the pre-test screening was failed four times, the program would present 30 trials at the lowest density to obtain a robust percent-correct score. However, during testing, this did not occur for any participant.

Three of the AB participants (AB1, AB6, and AB25) were tested during lab visits and used a lab-owned Naida Q90 and Connect, loaded with a copy of their clinical map. All AB participants used the Naida Connect except for AB2 and AB26, who connected using the ComPilot. Mixing ratios between the microphone and direct-audio connection were not obtained but were likely to be 70%/30%, based on previous clinical maps obtained from listeners. All but one user of a device made by Cochlear used a streamed audio connection from their Apple iPhone to their processor, using the made-for-iPhone (MFi) protocol and the Cochlear Phoneclip. Cochlear also provides the MiniMic 2+ to perform a similar function.

Early participant reports during piloting using the Compilot/Naida Connect suggested that the processor reverted to the microphone input if the wireless audio input did not transmit for a few seconds, and the first 0.5–1 s of the next sound transmitted to the processor would not be presented to the listener. Investigations using a processor connected to an oscilloscope confirmed this. To solve this problem, during silent portions of the test (duration >2 s), for example, when the participant responded to a trial, a silent, 30-s mp3 audio file was transmitted (and looped if necessary) until the next sound was presented, thus, maintaining the audio connection and preventing the processor from returning to mic-only input.

Because webSTRIPES uses a smaller number of reversals than loudspeaker STRIPES, we first evaluated the effect of the number of reversals on thresholds using the loudspeaker STRIPES data reported by Archer-Boyd et al. (2020). Figure 2 shows the correlation [Fig. 2(a)] and Bland–Altman plot (Bland and Altman, 1986) [Fig. 2(b)] between the loudspeaker STRIPES thresholds when calculated from the mean of the last four reversals (reversals 9–12, long version) and the penultimate four (reversals 5–8, short version). The across-listener Pearson correlation between the short and long versions is strong [r = 0.98, p < 0.001, degrees of freedom (df ) = 45]. Bland and Altman recommend first plotting the difference between the two sessions as a function of their mean [Fig. 2(b)]. This shows a moderate negative correlation (r = −0.49, p < 0.001, df = 45). Higher performers' thresholds improve if the adaptive track is longer (e.g., more reversals). The bias remains small (−0.3), and all except three points fall within ±1.96 s.d., which is indicated by the dashed lines. Bland and Altman suggested that 95% of points should lie within these bounds, and our results show 94% of points within the dotted lines, and two of the outliers are within density = 0.05 of the line. A two-way repeated-measures analysis of variance (ANOVA) using results from the loudspeaker-presentation conditions in Archer-Boyd et al. (2020), with factors “number of reversals” and “processing strategy” (HiRes vs Optima) revealed that fewer reversals produced significantly lower (worse) thresholds [F(1,9) = 5.474, p = 0.04].

Fig. 2.

Correlation and Bland–Altman plots for individual-run thresholds calculated from eight and 12 reversals. Solid and dashed lines in (b) show the mean and 1.96 times the standard deviation (s.d.) of the threshold difference.

Fig. 2.

Correlation and Bland–Altman plots for individual-run thresholds calculated from eight and 12 reversals. Solid and dashed lines in (b) show the mean and 1.96 times the standard deviation (s.d.) of the threshold difference.

Close modal

Analysis of adaptive staircase lengths showed that, on average, terminating a staircase at eight reversals would result in 26.8% fewer trials than the 12-reversal staircase (41.81 vs 57.17, respectively).

Figure 3 compares the webSTRIPES results with the loudspeaker STRIPES thresholds from Archer-Boyd et al. (2020) using the participants' clinical Optima processing strategy (Clin. Optima LS) and also shows the webSTRIPES results with Cochlear users using the ACE strategy. A paired t-test showed no effect of test type [t(8) = 1.97, p = 0.081] in the AB group. Therefore, there is no indication that changing from a 12-reversal lab-based loudspeaker presentation to a shorter, eight-reversal web-based presentation had an effect on performance overall. However, the performance of some participants (e.g., AB24, AB25, and AB26) does benefit from the longer, lab-based loudspeaker version. The thresholds of Cochlear users were not significantly different from the AB users [t(13) = 0.85, p = 0.41]. In addition, an analysis of the s.d. of the last four reversals in each run of webSTRIPES showed that mean s.d. for AB users was 0.36 (range: 0.12–1.4), similar to the 0.37 (range: 0.12–2.14) for Cochlear users.

Fig. 3.

Blue bars show the webSTRIPES results for AB (denoted as “AB#”) and Cochlear (denoted as “C#”) participants. Orange bars show the previous loudspeaker STRIPES results for AB participants. Squares show individual thresholds, and triangles show the follow-up sessions for AB24, described in Archer-Boyd and Carlyon (2019).

Fig. 3.

Blue bars show the webSTRIPES results for AB (denoted as “AB#”) and Cochlear (denoted as “C#”) participants. Orange bars show the previous loudspeaker STRIPES results for AB participants. Squares show individual thresholds, and triangles show the follow-up sessions for AB24, described in Archer-Boyd and Carlyon (2019).

Close modal

Figure 4 shows the correlation (a) and Bland–Altman plot (b) between the loudspeaker STRIPES thresholds (12 reversals) and the webSTRIPES results. The results are highly correlated across participants (r = 0.93, p < 0.001, df = 7). The Bland–Altman plot shows that the difference between thresholds as performance increases is negatively correlated (r = −0.8, p = 0.008, df = 7) and negatively biased by density = −0.99. Overall, this means that performance on webSTRIPES was comparable to loudspeaker STRIPES, with a trend toward larger differences in scores for high performers (AB24 and AB26).

Fig. 4.

Correlation and Bland–Altman plots for loudspeaker vs webSTRIPES. Solid and dashed lines in (b) show the mean and 1.96 times the s.d. of the threshold difference.

Fig. 4.

Correlation and Bland–Altman plots for loudspeaker vs webSTRIPES. Solid and dashed lines in (b) show the mean and 1.96 times the s.d. of the threshold difference.

Close modal

As shown in Sec. 4, there was a strong significant correlation between loudspeaker STRIPES thresholds calculated using 12 reversals (long version) and eight reversals (short version); although performance on the short version was reduced at high thresholds, most differences in threshold were inside the ±1.96 s.d. specified by Bland and Altman. The short version needed approximately 25% fewer trials to converge, leading to a time saving (important for a clinical test) of several minutes (approximately 20% of the total test time). The exact time saving for any one participant and condition depended on a number of factors, including the threshold density, the variability in densities presented to the listener, and the speed of convergence after the eighth reversal (i.e., how close together the subsequent reversals were). On average, the whole test, including pre-tests and training, took about 12.5 min with a range from about 8 min to (for AB26, who had very high thresholds) nearly 17 min.

No significant difference was found in overall performance between the loudspeaker STRIPES (obtained with each listener's clinical Optima strategy) and webSTRIPES thresholds across nine AB participants. The correlation analysis showed a negative correlation between mean threshold and difference between thresholds for loudspeaker STRIPES and webSTRIPES. However, this correlation was driven by the results of three participants, including listeners AB24 and AB26, who showed the highest and second-highest thresholds (best performance) of all participants in the loudspeaker STRIPES condition. These thresholds were also substantially higher than their loudspeaker threshold obtained with a HiRes strategy in our previous study, leading to the suggestion that these high STRIPES densities were achieved by latching on to some extraneous cue that was specific to the Optima S strategy for stimuli presented over loudspeakers (Archer-Boyd et al., 2020). It is possible that this cue was not available in the WebSTRIPES implementation, either because of the stimulus presentation and/or generation method or because of the increased auditory and/or visual distraction arising from performing the test outside of the laboratory.

An outstanding question in the STRIPES literature to date was its compatibility with non-AB devices and devices that use N-of-M strategies instead of continuous interleaved sampling (CIS). This concern was driven by the AB-centred design of the original STRIPES test (Archer-Boyd et al., 2018) and a lack of published evidence. The current study provides that (limited) evidence. Thresholds were not significantly different between the Cochlear and AB groups, and the s.d. of the last four reversals was broadly similar for users of the two devices. It might have been expected that the large differences in stimulation method between AB and Cochlear devices (e.g., CIS vs N-of-M, very different filter roll-off rates) may have produced different thresholds. However, this does not occur, consistent with the findings of many studies of speech intelligibility that compare different manufacturers. STRIPES stimuli contain much more redundant information than speech. In addition, visual inspection of STRIPES stimulus electrodograms produced using CIS and N-of-M strategies show very little differences. This is because N-of-M strategies such as ACE are normally set to select eight peaks in each processing frame and to stimulate the eight corresponding electrodes. STRIPES stimuli are very sparse by design, meaning that eight peaks are sufficient to represent the whole stimulus at any instant up to very high densities, and the end result appears to be similar to CIS. The similarity of thresholds between AB and Cochlear users suggests that the test will also work with Med-El users, although further research is required to confirm this.

CI users as a group hold a specific advantage for online testing over other cohorts. Since there are only between one and three devices that can be used to stream audio to each manufacturer (often using the same audio signal path in a different hardware package), the signal path and its effects on stimuli can be estimated almost entirely before deployment. However, there remain some limitations. There is no guarantee that participants will use a wireless device to connect to their phone/tablet/laptop or necessarily notice that the sound is being produced by the device, rather than transmitted directly to the processor. In this study, all participants had been recruited previously, and we trusted them to follow the instructions. In addition, if they had issues connecting their devices or had not used them, we invited them to the lab and performed the test there. Screening methods for acoustic-only hearing participants that check whether they are using headphones (Woods et al., 2017) are not generally feasible for CI listeners as they require binaural hearing. We are therefore developing an alternative test that we hope to incorporate into a later version. Another limitation is that the “mixing ratio” is set by the audiologist, and it appears that the microphone is never set to 0% for safety and environmental monitoring reasons. Therefore, experimenters do not know how much environmental noise is audible to the participant. This issue can be partially mitigated by instructing participants to find a quiet place to perform the task and including a comfortable-level check at the beginning of the test to make sure the stimuli will be sufficiently audible to the listener. Listeners may not have the ability to stream audio from their phone/tablet/laptop, either because they do not have the correct peripheral or have it but do not feel the need to use it. For this reason, three of the participants performed the task during a visit to the lab. Therefore, the results presented here are a proof-of-concept rather than a fully ecologically valid study. However, new generations of CI devices, such as, for example, AB's MarvelTM processor, connect directly to streaming devices without the need for another device as an intermediary. Cochlear CI devices have also been capable of this with Apple devices for a number of years and do now also provide this functionality for compatible Android devices. Hence, as CI devices develop and CI users obtain an updated speech processor, compatibility and ease-of-use will improve.

The effect of reducing the number of reversals and presenting the STRIPES test via the web and wireless audio streaming was investigated. Previous loudspeaker STRIPES results were compared to results obtained from the same listeners using the new online webSTRIPES version of the test. Cochlear device users were also tested for the first time with webSTRIPES. Analysing the loudspeaker results, no effect of reducing the number of reversals was found across listeners. No effect of presentation type (loudspeaker vs online) was found, and results obtained with the Cochlear device converged on thresholds in a stable manner and were broadly comparable to those obtained with AB listeners.

This work was supported by Medical Research Council Grant No. G101400 to R.P.C. We thank John Deeks for his help and advice. The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Information on how to download and use the test can be obtained from https://www.hearing-research.group.cam.ac.uk/software/.

1.
Archer-Boyd
,
A. W.
, and
Carlyon
,
R. P.
(
2019
). “
Simulations of the effect of unlinked cochlear-implant automatic gain control and head movement on interaural level differences
,”
J. Acoust. Soc. Am.
145
,
1389
1400
.
2.
Archer-Boyd
,
A. W.
,
Goehring
,
T.
, and
Carlyon
,
R. P.
(
2020
). “
The effect of free-field presentation and processing strategy on a measure of spectro-temporal processing by cochlear-implant listeners
,”
Trends Hear.
24
,
2331216520964281
.
3.
Archer-Boyd
,
A. W.
,
Southwell
,
R. V.
,
Deeks
,
J. M.
,
Turner
,
R. E.
, and
Carlyon
,
R. P.
(
2018
). “
Development and validation of a spectro-temporal processing test for cochlear-implant users
,”
J. Acoust. Soc. Am.
144
,
2983
2997
.
4.
Aronoff
,
J. M.
, and
Landsberger
,
D. M.
(
2013
). “
The development of a modified spectral ripple test
,”
J. Acoust. Soc. Am.
134
,
EL217
EL222
.
5.
Azadpour
,
M.
, and
McKay
,
C. M.
(
2012
). “
A psychophysical method for measuring spatial resolution in cochlear implants
,”
J. Assoc. Res. Otolaryngol.
13
,
145
157
.
6.
Berg
,
K. A.
,
Noble
,
J. H.
,
Dawant
,
B. M.
,
Dwyer
,
R. T.
,
Labadie
,
R. F.
, and
Gifford
,
R. H.
(
2019
). “
Speech recognition as a function of the number of channels in perimodiolar electrode recipients
,”
J. Acoust. Soc. Am.
145
,
1556
1564
.
7.
Bland
,
J. M.
, and
Altman
,
D. G.
(
1986
). “
Statistical methods for assessing agreement between two methods of clinical measurement
,”
Lancet
327
,
307
310
.
8.
Carlyon
,
R. P.
, and
Goehring
,
T.
(
2021
). “
Cochlear implant research and development in the twenty-first century: A critical update
,”
J. Assoc. Res. Otolaryngol.
22
,
481
508
.
9.
Croghan
,
N. B. H.
,
Duran
,
S. I.
, and
Smith
,
Z. M.
(
2017
). “
Re-examining the relationship between number of cochlear implant channels and maximal speech intelligibility
,”
J. Acoust. Soc. Am.
142
,
EL537
EL543
.
10.
Davis
,
M. H.
, and
Johnsrude
,
I. S.
(
2007
). “
Hearing speech sounds: Top-down influences on the interface between audition and speech perception
,”
Hear. Res.
229
,
132
147
.
11.
Davis
,
M. H.
,
Johnsrude
,
I. S.
,
Hervais-Adelman
,
A.
,
Taylor
,
K.
, and
McGettigan
,
C.
(
2005
). “
Lexical information drives perceptual learning of distorted speech: Evidence from the comprehension of noise-vocoded sentences
,”
J. Exp. Psychol. Gen.
134
,
222
241
.
12.
De Graaff
,
F.
,
Huysmans
,
E.
,
Merkus
,
P.
,
Theo Goverts
,
S.
, and
Smits
,
C.
(
2018
). “
Assessment of speech recognition abilities in quiet and in noise: A comparison between self-administered home testing and testing in the clinic for adult cochlear implant users
,”
Int. J. Audiol.
57
,
872
880
.
13.
Drennan
,
W. R.
,
Longnion
,
J. K.
,
Ruffin
,
C.
, and
Rubinstein
,
J. T.
(
2008
). “
Discrimination of Schroeder-phase harmonic complexes by normal-hearing and cochlear-implant listeners
,”
J. Assoc. Res. Otolaryngol.
9
,
138
149
.
14.
Goehring
,
J. L.
,
Hughes
,
M. L.
,
Baudhuin
,
J. L.
,
Valente
,
D. L.
,
McCreery
,
R. W.
,
Diaz
,
G. R.
,
Sanford
,
T.
, and
Harpster
,
R.
(
2012
). “
The effect of technology and testing environment on speech perception using telehealth with cochlear implant recipients
,”
J. Speech Lang. Hear. Res.
55
(
5
),
1373
1386
.
15.
Goehring
,
T.
,
Archer-Boyd
,
A.
,
Deeks
,
J. M.
,
Arenberg
,
J. G.
, and
Carlyon
,
R. P.
(
2019
). “
A site-selection strategy based on polarity sensitivity for cochlear implants: Effects on spectro-temporal resolution and speech perception
,”
J. Assoc. Res. Otolaryngol.
20
,
431
448
.
16.
Goehring
,
T.
,
Arenberg
,
J. G.
, and
Carlyon
,
R. P.
(
2020
). “
Using spectral blurring to assess effects of channel interaction on speech-in-noise perception with cochlear implants
,”
J. Assoc. Res. Otolaryngol.
21
(
4
),
353
371
.
17.
Henry
,
B. A.
, and
Turner
,
C. W.
(
2003
). “
The resolution of complex spectral patterns by cochlear implant and normal-hearing listeners
,”
J. Acoust. Soc. Am.
113
,
2861
2873
.
18.
Lange
,
K.
,
Kühn
,
S.
, and
Filevich
,
E.
(
2015
). “‘
Just another tool for online studies’ (JATOS): An easy solution for setup and management of web servers supporting online studies
,”
PLoS One
10
,
e0130834
.
19.
Litvak
,
L. M.
,
Spahr
,
A. J.
,
Saoji
,
A. A.
, and
Fridman
,
G. Y.
(
2007
). “
Relationship between perception of spectral ripple and speech recognition in cochlear implant and vocoder listeners
,”
J. Acoust. Soc. Am.
122
,
982
991
.
20.
MathWorks
(
2019
).
MATLAB 2019a
(
MathWorks Inc
.,
Natick, MA
).
21.
Peng
,
Z. E.
,
Buss
,
E.
,
Shen
,
Y.
,
Bharadwaj
,
H.
,
Stecker
,
G. C.
,
Beim
,
J. A.
,
Bosen
,
A. K.
,
Braza
,
M.
,
Diedesch
,
A. C.
, and
Dorey
,
C. M.
(
2020
). “
Remote testing for psychological and physiological acoustics: Initial report of the P&P task force on remote testing
,”
Proc. Mtgs. Acoust.
42
,
050009
.
22.
Saoji
,
A. A.
,
Litvak
,
L.
,
Spahr
,
A. J.
, and
Eddins
,
D. A.
(
2009
). “
Spectral modulation detection and vowel and consonant identifications in cochlear implant listeners
,”
J. Acoust. Soc. Am.
126
,
955
958
.
23.
Schvartz-Leyzac
,
K. C.
,
Zwolan
,
T. A.
, and
Pfingst
,
B. E.
(
2017
). “
Effects of electrode deactivation on speech recognition in multichannel cochlear implant recipients
,”
Cochlear Implants Int.
18
,
324
334
.
24.
Supin
,
A.
,
Popov
,
V. V.
,
Milekhina
,
O. N.
, and
Tarakanov
,
M. B.
(
1994
). “
Frequency resolving power measured by rippled noise
,”
Hear. Res.
78
,
31
40
.
25.
Won
,
J. H.
,
Drennan
,
W. R.
,
Nie
,
K.
,
Jameyson
,
E. M.
, and
Rubinstein
,
J. T.
(
2011
). “
Acoustic temporal modulation detection and speech perception in cochlear implant listeners
,”
J. Acoust. Soc. Am.
130
,
376
388
.
26.
Won
,
J. H.
,
Drennan
,
W. R.
, and
Rubinstein
,
J. T.
(
2007
). “
Spectral-ripple resolution correlates with speech reception in noise in cochlear implant users
,”
J. Assoc. Res. Otolaryngol.
8
,
384
392
.
27.
Woods
,
K. J.
,
Siegel
,
M. H.
,
Traer
,
J.
, and
McDermott
,
J. H.
(
2017
). “
Headphone screening to facilitate web-based auditory experiments
,”
Atten. Percept. Psychophys.
79
,
2064
2072
.