Related Experiment Videos
Vowel formant discrimination for high-fidelity speech
1Department of Speech and Hearing Sciences, Indiana University, Bloomington, Indiana 47405, USA. cliu5@buffalo.edu
The Journal of the Acoustical Society of America
|September 21, 2004
Summary
Normal-hearing listeners require 0.37 barks to discriminate vowel formants in sentences. Performance was best in word-final positions, suggesting context influences vowel perception.
Area of Science:
- Acoustics
- Speech Perception
- Psychoacoustics
Background:
- Vowel formant discrimination is crucial for speech intelligibility.
- Understanding how listeners perceive vowels in naturalistic speech is essential.
Purpose of the Study:
- To determine normal-hearing listeners' ability to discriminate vowel formants in everyday speech.
- To investigate the influence of phonetic context and word position on vowel discrimination.
Main Methods:
- Vowel formant discrimination thresholds (F1 and F2) were measured for /I, epsilon, ae, lambda/ in /bVd/ syllables.
- Speech stimuli were synthesized using the high-fidelity STRAIGHT method.
- Phonetic context (syllables, phrases, sentences) and word position were manipulated.
Main Results:
- Discrimination thresholds were not significantly affected by longer phonetic context or the addition of an identification task.
- Word-final positions showed significantly better discrimination performance compared to initial or middle positions.
- An average of 0.37 barks was required for discrimination in sentences, an 84% increase from isolated vowels.
Conclusions:
- Normal-hearing listeners require more precise vowel formant information in sentences than in isolated vowels.
- Word position significantly impacts vowel formant discrimination, with final positions being easier.
- High-fidelity speech synthesis (STRAIGHT) may present greater spectral-temporal variability, slightly elevating discrimination thresholds compared to traditional methods.