The role of periodicity in the perception of masked speech with simulated and real cochlear implants

Transcription

1 The role of periodicity in the perception of masked speech with simulated and real cochlear implants Kurt Steinmetzger and Stuart Rosen UCL Speech, Hearing and Phonetic Sciences Heidelberg, 09. November

2 CI users stimuli In normal hearing, periodic sounds are much less effective maskers of speech than aperiodic ones (Steinmetzger and Rosen, 2015). For various reasons... To what extent is this benefit preserved in simulated and real CIs? If the benefit is reduced, this helps explain why CI users face such difficulties when attempting to understand speech in noisy environments. The annoying student asks too many questions. 2

3 CI users stimuli Normal-hearing listeners also benefit substantially from slow modulations of the masker amplitude ( glimpsing ) This effect is usually strongly reduced or absent in CI users. But why exactly? Maskers that promote glimpsing (+MR; Kwon et al., 2012) to test whether this is due to masking or the inability to segregate streams The annoying student asks too many questions. 3

4 CI simulations signal processing Signal processing steps involved in CI simulations Noise-vocoding is with 8 channels and a 400 Hz envelope low-pass Standard simulation that preserves temporal periodicity cues 4

5 CI simulations stimuli Methods & procedure: 11 NH listeners, 18 conditions, 20 sentences per condition Adaptive procedure determining SNR needed to correctly perceive 50% of keywords ( SRT50 ) Mixed-model regression statistics 5

6 CI simulations results Significant main effects of masker periodicity and masker envelope (both p < 0.001), nothing else In summary: Masker-periodicity is reduced but still present (~3.5 db). Same performance with steady and +MR maskers! Likely explanation in both cases: different modulation spectra 6

7 CI simulations stimulus modulations Modulation spectrograms based on front end output of the mr-sepsm speech intelligibility model (Jørgensen et al., 2013) Shows the modulation power for each combination of auditory and modulation filter, for each stimulus condition 7

8 CI users participant information Subject Age Sex Age at onset of deafness Years of implant use Aetiology of deafness Implant fitting Implant type (processing strategy) 1 70 M 45 2 Sensorineural Right CI512 (ACE) 2 69 F 53 3 Ménière s Right CI422 (ACE) 3 82 F 70 3 Unknown Right CI422 (ACE) 4 65 F 38 9 Unknown Left HiRes 90K (HiRes Optima) 5 60 F 25 2 Unknown Left CI512 (ACE) 6 49 F 23 2 Sensorineural Right HiRes 90K Adv. (HiRes Optima) 7 75 F 35 3/3 Hereditary Both CI422 (ACE) & CI422 (ACE) 8 73 F 50 13/11 Ménière s Both CI24R (ACE) & CI24RE (ACE) 8

9 CI users results Same procedure as before, but only 12 conditions to make experiment less straining Small (~1.2 db) but significant main effects of masker periodicity (p < 0.01). Strong interaction of +MR maskers and target periodicity (p < 0.001). SRTs ~2.8 db worse if target is aperiodic 9

10 General discussion 1. Why is masker-periodicity benefit almost absent in CI users? CIs don't transmit random envelope fluctuations (Oxenham & Kreft, 2014). Thus, only pitch-related envelope modulations left as cue. 2. No glimpsing even if there is hardly any energetic overlap between target speech and masker (+MR): Inability to segregate streams seems to be the main issue, especially as target and masker have similar slow envelope modulations. 3. CI users perform markedly worse with the +MR maskers if there are no periodicity cues in the target speech: Due to spectral smearing, these periodicity cues are the only acoustic feature they can exploit. This may make them more salient, or even draw their attention. 10

11 References Jørgensen, S., Ewert, S. D., and Dau, T. (2013). "A multi-resolution envelope-power based model for speech intelligibility," J. Acoust. Soc. Am. 134, Kwon, B. J., Perry, T. T., Wilhelm, C. L., and Healy, E. W. (2012). "Sentence recognition in noise promoting or suppressing masking release by normal-hearing and cochlear-implant listeners," J. Acoust. Soc. Am. 131, Oxenham, A. J., and Kreft, H. A. (2014). "Speech perception in tones and noise via cochlear implants reveals influence of spectral resolution on temporal processing," Trends Hear. 18, Steinmetzger, K., and Rosen, S. (2015). "The role of periodicity in perceiving speech in quiet and in background noise," J. Acoust. Soc. Am. 138,