Does acoustic fundamental frequency information enhance cochlear implant

  • Slides: 1
Download presentation
Does acoustic fundamental frequency information enhance cochlear implant performance? Laura Mulhern and Dr. Helen

Does acoustic fundamental frequency information enhance cochlear implant performance? Laura Mulhern and Dr. Helen Cullington lmm 204@soton. ac. uk INTRODUCTION Low frequency information via a hearing aid has been shown to increase speech intelligibility in noise for CI listeners. Some studies suggest fundamental frequency (F 0) provides this advantage (Brown and Bacon, 2009; Chang et al. , 2006), while others question the importance of F 0 (Kong and Carlyon, 2007). Currently one study assesses the benefit with CI patients (Brown and Bacon, 2009); the others use CI simulations. They found a significant benefit of acoustic F 0 information to CI listeners, however: 1. they used patients who have exceptionally good residual hearing 2. They only provide the target acoustically, thus increasing the signal to noise ratio (SNR) HYPOTHESES DISCUSSION 1. If the addition of F 0 provides significant benefit to speech recognition in noise then F 0 information must provide a cue to aid speech in noise performance. 2. If the addition of F 0 provides equal benefit to the addition of the entire acoustic information with all other cues omitted, the F 0 must be the salient cue that provides the benefit. RESULTS A one-way repeated measures ANOVA showed no significant effect on speech intelligibility in noise between the three listening conditions (p=. 126). Our results suggest that the acoustic low-frequency and F 0 information have no significant impact on speech recognition in noise in CI listeners. • This conclusion is not in agreement with other studies that suggest that low-frequency acoustic cues aid speech recognition (Chang et al. , 2006) or that acoustic F 0 cues aid CI listeners (Brown and Bacon, 2009). • Possible reasons for the discrepancy: 1. both the target and masker were presented acoustically (therefore the SNR was not changed) 2. typical CI listeners were used (those who had any amount of residual hearing below 500 Hz contralateral to the CI) not good ‘EAS’ users. • Despite no statistical difference, most participants have chosen to continue wearing a hearing aid since they feel they obtain a benefit. • One participant (F 0 Filt 7, not recorded in results) was unable to achieve an adequate SRT score for the CI alone condition yet could successfully complete the adaptive test for both the filt and F 0 conditions. METHODOLOGY Aim To investigate the contribution of acoustic lowfrequency and F 0 information using typical CI users. Limitations Participants • Seven adult participants aged 34 -84 years, with a score >75% on BKB sentences in quiet and aidable residual hearing <500 Hz. Low frequency acoustic information may have been too quiet for the subjects to obtain benefit. • Only ‘good’ CI users participated (score >75% on BKB sentences in quiet) due to the use of the adaptive test in noise. • Low statistical power (5%). Table 1: Demographic data for participants Materials Figure 2: Individual SRTs for all seven participants in the three listening conditions. Colours used to show results in Figure 2 are matched to participants in Table 1. Target: BKB sentences spoken by male Masker: IEEE sentences spoken by female Acoustic target and masker filtered through MATLAB using 3 rd order elliptical filters with a cutoff at 500 Hz. F 0 of target and masker extracted using PRAAT speech synthesis program. CONCLUSIONS • Low-frequency acoustic sound (<500 Hz) does not benefit speech recognition in noise for the participants in this study. • F 0 acoustic information does not benefit speech recognition in noise for the participants in this study. FUTURE DIRECTION The experiment should be repeated with a larger sample size, thus improving the statistical power. Unfiltered speech presented through a loudspeaker (target constant at 60 d. B (A)). Filtered speech presented contralaterally via an insert earphone, intensity level adjusted to an audible and comfortable level for each participant. LITERATURE Brown, C. A. and Bacon, S. P. 2009. Achieving electricacoustic benefit with modulated tone. Ear and Hearing. (In press). Figure 1: A spectrogram of an example sentence highlighting the extracted F 0. The blue line indicates the F 0, that has a mean frequency of 151 Hz and inter-sentence variation from 98 – 300 Hz. Design 1 -up, 1 -down adaptive procedure to measure the speech reception threshold (SRT) of three listening conditions: 1. CI alone 2. CI and acoustic <500 Hz filtered speech contralaterally (filt) 3. CI and acoustic F 0 filtered speech contralaterally (F 0) SRT: SNR at which subjects score 50% correct Figure 3: The median, minimum and maximum data values and the interquartile range of SRT scores for each listening condition. No correlations were found between filt acoustic benefit (CI SRT minus filt SRT) or F 0 acoustic benefit (CI SRT minus F 0 SRT) and participant age, gender, processor, length of CI experience, residual hearing or bimodal status. A one-way repeated measures ANOVA on the order of conditions was insignificant, therefore any learning effect was negligible. Chang, J. E. , Bai, J. Y. and Zeng, F. 2006. Unintelligible lowfrequency sound enhances simulated cochlear-implant speech recognition in noise. IEEE Transactions on Biomedical Engineering 53, 2598 -2601. Cullington, H. E. and Zeng, F. 2009. Bimodal hearing benefit for speech recognition with competing voice in cochlear implant subject with normal hearing in contralateral ear. Ear and Hearing. (In press). Kong, Y. and Carlyon, R. P. 2007. Improved speech recognition in noise in simulated binaurally combined acoustic and electric stimulation. The Journal of the Acoustical Society of America 121, 3717 -3727. ACKNOWLEDGEMENTS Patients and staff at the South of England Cochlear Implant Centre