Evaluation of a novel phrase-based speech-recognition test using synthetic speech

Area of Science:

• Audiology
• Speech Science
• Psychoacoustics

Background:

• Synthesized speech tests are crucial for audiological evaluations.
• The Oldenburg Phrases (OLPHRA) test offers a structured format for speech recognition assessment.
• Standardized and validated test lists are essential for reliable audiological outcomes.

Purpose of the Study:

• To compile and evaluate test lists for the Oldenburg Phrases (OLPHRA) synthesized speech test.
• To assess the perceptual and acoustical equivalence of OLPHRA test lists.
• To compare OLPHRA with established speech tests and evaluate its suitability for audiology.

Main Methods:

• Development of 1000 phrase lists based on properties like speech-recognition thresholds (SRT) and phoneme classes.
• Random selection of 21 lists from 1000 for analysis of perceptual equivalence and training effects.
• Comparison of OLPHRA with established speech tests in young normal-hearing (YNH) and older hearing-impaired (OHI) participants.

Main Results:

• 21 OLPHRA lists demonstrated perceptual equivalence in YNH participants (average SRT -9.0 dB SNR, max difference 1.2 dB SNR).
• Acoustical equivalence was confirmed by the Speech Intelligibility Index; phoneme class equivalence was also observed.
• OLPHRA showed no significant SRT difference compared to established tests in OHI participants, with high test-retest reliability (ICC ≥ 0.9).

Conclusions:

• OLPHRA test lists are perceptually and acoustically equivalent, fulfilling ISO 8253-3 requirements.
• The test exhibits minimal training effects and is suitable for diverse audiological applications.
• OLPHRA provides a versatile and reliable tool for audiology, comparable to existing speech tests.