Improving Speech Recognition with Real-Time Contrast Enhancement
Author Information
Author(s): Alexander Joshua M., Jenison Rick L., Kluender Keith R.
Primary Institution: Purdue University and University of Wisconsin-Madison
Hypothesis
Can a real-time signal-processing algorithm enhance speech recognition for normal-hearing listeners?
Conclusion
The Contrast Enhancement algorithm significantly improved consonant and vowel identification in both quiet and noisy conditions.
Supporting Evidence
- The CE algorithm improved consonant identification rates significantly in both quiet and noisy conditions.
- Listeners showed consistent improvement in identifying place of articulation for consonants.
- The algorithm was effective regardless of the degree of spectral smearing.
Takeaway
This study shows that a special computer program can help people hear speech better, especially when it's noisy.
Methodology
Normal-hearing listeners identified consonants and vowels processed by the Contrast Enhancement algorithm in quiet and noisy environments.
Limitations
The study used normal-hearing listeners to simulate hearing loss, which may not fully represent actual hearing-impaired individuals.
Participant Demographics
Undergraduate students from the University of Wisconsin-Madison, all native speakers of American English with normal hearing.
Statistical Information
P-Value
p<0.001
Statistical Significance
p<0.001
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website