Acoustic analysis and detection of hypernasality using a group delay function

Area of Science:

• Speech processing
• Acoustic analysis
• Speech pathology

Background:

• Hypernasality, a speech disorder, results from abnormal airflow through the nasal cavity.
• Traditional acoustic analysis faces challenges in resolving closely spaced formants characteristic of hypernasal speech.
• A consistent low-frequency resonance around 250 Hz is observed in hypernasal speech.

Purpose of the Study:

• To develop and validate a novel group delay-based signal processing technique for hypernasality detection.
• To introduce a new acoustic measure derived from the band-limited group delay spectrum.
• To assess the effectiveness of the proposed measure across different vowel phonemes.

Main Methods:

• A group delay-based signal processing technique was employed.
• A band-limited approach was developed to estimate formant locations in hypernasal speech.
• A new acoustic measure was defined using the band-limited group delay spectrum.

Main Results:

• The proposed acoustic measure achieved 100% detection accuracy for the phoneme /a/.
• Detection accuracies for /i/ and /u/ were 88.78% and 86.66%, respectively.
• The method's effectiveness was validated on diverse speech data from different recording environments.

Conclusions:

• The group delay-based acoustic measure is a promising tool for accurate hypernasality detection.
• The low-frequency acoustic characteristics of hypernasal speech are key indicators for detection.
• The developed technique offers a robust solution for analyzing and identifying hypernasal speech patterns.