Real-Time Signal Processing in Speech Recognition and Its Potential Use within The Development of Hearing Aids

Area of Science:

• Speech Processing
• Acoustic Phonetics
• Signal Processing

Background:

• Rule-based speech recognition requires robust acoustic-phonetic features.
• Speech signals are susceptible to dynamic variability and environmental noise.
• Real-time implementation is crucial for practical speech recognition systems.

Purpose of the Study:

• To describe essential acoustic-phonetic features for rule-based speech recognition.
• To evaluate feature algorithms based on robustness, noise resilience, and real-time suitability.
• To explore the application of these features in developing advanced hearing aids.

Main Methods:

• Feature selection based on robustness to speech dynamics and environmental noise.
• Real-time estimation of features using a 32-bit floating-point digital signal processor (DSP32).
• Analysis of acoustic-phonetic features: pitch, formants, segmentation, and labelling.

Main Results:

• Identified key acoustic-phonetic features suitable for real-time speech recognition.
• Demonstrated the feasibility of real-time feature estimation using DSP32.
• Highlighted the potential utility of these features for future hearing aid research.

Conclusions:

• Acoustic-phonetic features are vital for effective rule-based speech recognition.
• The described features offer robustness and real-time processing capabilities.
• These features hold promise for enhancing the functionality of future hearing aid technologies.