200 likes | 324 Views
Evaluation of a Binaural FMV Beamforming Algorithm in Noise. Jeffery B. Larsen, Charissa R. Lansing, Robert C. Bilger, Bruce Wheeler, Sandeep Phatak, Nandini Iyer, Mike Lockwood, William O’Brien, Doug Jones, and Albert S. Feng University of Illinois Department of Speech and Hearing Sciences
E N D
Evaluation of a Binaural FMV Beamforming Algorithm in Noise Jeffery B. Larsen, Charissa R. Lansing, Robert C. Bilger, Bruce Wheeler, Sandeep Phatak, Nandini Iyer, Mike Lockwood, William O’Brien, Doug Jones, and Albert S. Feng University of Illinois Department of Speech and Hearing Sciences Beckman Institute Research supported by NIH grant #DC04840
INTRODUCTION • A Frequency Banded Minimum Variance (FMV) algorithm has been developed • Preliminary investigations with the FMV in simulation have been promising • Evaluation of the FMV in a multi-source environment is needed to demonstrate its effectiveness • The primary question of this study is: How much benefit from the FMV is obtained in noisy environments beyond that provided by directional microphones alone?
GOALS OF THE CURRENT STUDY • Determine if the positive results with FMV algorithm are maintained in a true multi-source environment with competing speech • Quantify the SNR benefit of the FMV algorithm beyond that obtained from directional microphones alone • Obtain data to validate improved intelligibility with FMV algorithm in noise over directional microphones alone • Assess FMV performance in multiple jammer configurations and noise types
Setup of current preliminary investigation - Experiment #1 • Dependent variables • SRT in quiet and in noise with spondaic words • SRT in quiet and in noise with HINT sentences (Nilsson, Soli, & Sullivan, 1994) • Subject rated intelligibility of SIR passages (Cox & McDaniel, 1989) at a -4 dB SNR • 5 listeners w/ NH (20 - 40 years old) and 9 w/ SNHL (53 to 81 years old)
CD player Setup #1 * * Target * Computer Cardioid mics * * Reversed speech of four talkers speaking R-SPIN sentences at -40°, -20°, 20°, and 40° from target at 0° Real-time system Probe mic system
Setup of current preliminary investigation - Experiment #2 • Dependent variables • SRT in quiet and in noise with HINT sentences (Nilsson, Soli, & Sullivan, 1994) • Percent Correct of Keywords for CST Passages (Cox, Alexander, & Gilmore, 1987) at a 0 dB SNR • 5 listeners w/ SNHL (65 - 80 years old) • Use of speech-shaped cafeteria noise (Ricketts & Dhar, 1999) as jammer • Novel spacing and levels of jammers
CONCLUSIONS • FMV algorithm provides benefit across: • Competing signal types • Speech shaped cafeteria noise and reversed speech • ESSENTIALLY EQUAL BENEFIT IN EACH • Response Tasks • Intelligibility ratings - 50% higher for FMV than DM • Speech recognition - 36% improvement in WRS • Fixed and variable SRT measures • 6 TO 8 dB OF SRT IMPROVEMENT OVER DIRECTIONAL MICS ALONE FOR NORMAL HEARING LISTENERS • 8 TO 13 dB OF SRT IMPROVEMENT FOR LISTENERS WITH SNHL OVER DIRECTIONAL MICS ALONE DEPENDING UPON LOCATION OF JAMMERS
CONCLUSIONS (cont.) • Auditory Scenes • Setup #1 - jammers equal at +20º and +40º (Thanksgiving table scene) • VARIABLE SRT IMPROVEMENT OVER DIRECTIONAL MICS FOR LISTENERS WITH SNHL WAS 8 TO 11 dB • Setup #2 - Closest jammer at +20º was more intense than other three jammers at -80º, -40º, and +60º (4th of July picnic scene) • MEAN VARIABLE SRT IMPROVEMENT OVER DIRECTIONAL MICS FOR LISTENERS WITH SNHL WAS 13 dB (range of 8 to 19 dB across listeners)
FUTURE DIRECTIONS • Direct comparison of FMV with commercially available hearing aids with technology for listening in noise • Continued assessment of performance in different noise source configurations and reverberation times • Comparison of speech intelligibility performance with subjective measures of quality • Comparison of human performance with engineering metrics
Key References • Cox, R.M. & McDaniel, D.M. (1989). Development of the Speech Intelligibility Rating (SIR) Test for hearing aid comparisons. Journal of Speech, Language, and Hearing Research, 32(2), 347-352. • Cox, R.M., Alexander, G.C., & Gilmore, C. (1987). Development of the connected speech test (CST). Ear and Hearing, 8, 119S-126S. • Elledge, M.E ,Lockwood, M.E., Bilger, R.C., Goueygou, M., Jones, D.L., Lansing, C. R., Liu, C., O'Brien, W. D., Wheeler, B. C., & Feng, A. (1999).A real-time dual-microphone signal-processing system for hearingaids. Journal of the Acoustical Society of America,106(4), p.2279. • Lockwood, M.E., Jones, D.L., Elledge, M.E., Bilger, R.C., Goueygou, M., Lansing, C. R., Liu, C., O'Brien, W. D., Wheeler, B. C., & Feng, A. (1999). A minimum-variance frequency-domain algorithm for binaural hearingaid processing. Journal of the Acoustical Society of America,106(4), p.2278. • Nilsson, M.J., Soli, S.D., & Sullivan, J. (1994). Development of a hearing in noise test for the measurement of speech reception threshold. Journal of the Acoustical Society of America, 95, 1985-1999. • Ricketts, T., & Dahr, S. (1999). Comparison of performance across three directional hearing aids. Journal of the American Academy of Audiology, 10(4), 180-189.