Related Experiment Videos
Speaking rhythmically improves speech recognition under "cocktail-party" conditions
Mengyuan Wang1, Lingzhi Kong2, Changxin Zhang3
1Beijing Key Lab of Applied Experimental Psychology, School of Psychology, Beijing Normal University, Beijing 100875, China.
The Journal of the Acoustical Society of America
|May 3, 2018
Summary
Normal speech rhythm aids keyword recognition in noisy environments, especially against competing voices. This benefit disappears with unnatural, mixed-rhythm sentences, suggesting rhythm aids prediction.
Area of Science:
- Auditory neuroscience
- Speech perception
- Psychoacoustics
Background:
- The
Purpose of the Study:
- Investigate the impact of speech rhythm on speech recognition in challenging auditory conditions.
- Determine if consistent speech rhythm aids in processing speech amidst background noise.
Main Methods:
- Participants listened to target sentences with varying rhythmic structures (normal, hybrid) against different maskers (two-talker, speech-spectrum noise).
- Speech recognition accuracy was measured for keywords at different positions within the sentences.
Main Results:
- Keyword recognition improved for later words in normally rhythmic sentences against a two-talker masker.
- This positional benefit vanished for hybrid sentences with altered rhythms (fast, normal, slow).
- No significant positional effect was observed when using a speech-spectrum noise masker.
Conclusions:
- Consistent speech rhythm enhances speech recognition, particularly in the presence of informational masking like competing speech.
- The findings suggest that predictable, normal rhythm facilitates temporal prediction of target words, improving comprehension.
- Speech rhythm's role in auditory scene analysis and speech intelligibility is highlighted, especially in complex acoustic environments.