Robust cortical encoding of slow temporal modulations of speech

Area of Science:

• Neuroscience
• Auditory Neuroscience
• Speech Processing

Background:

• Understanding speech recognition in noisy environments is crucial.
• The brain's ability to isolate speech from background interference is not fully understood.
• Neural mechanisms underlying speech perception in complex acoustic scenes require further investigation.

Purpose of the Study:

• To investigate the neural representation of speech in challenging listening conditions.
• To determine how the brain synchronizes to speech envelopes amidst competing sounds.
• To identify neural correlates of speech intelligibility and background noise invariance.

Main Methods:

• Magnetoencephalography (MEG) was used to record brain activity.
• Subjects listened to speech masked by either competing speech or stationary noise.
• Cortical activity synchronization to the speech temporal envelope was analyzed.

Main Results:

• Neural synchronization to speech is robust against masking by competing speech (up to 8 dB) and noise (up to 9 dB).
• Precise neural synchronization to speech predicts subjective speech intelligibility in noise.
• Longer-latency neural responses (∼100 ms) exhibit speech selectivity and noise invariance, unlike shorter-latency responses (∼50 ms).

Conclusions:

• Neural synchronization to the speech envelope provides a mechanism for robust speech recognition in complex acoustic environments.
• A processing transition occurs in the auditory cortex, shifting from acoustic scene encoding to behaviorally relevant auditory object encoding.
• This neural mechanism is a strong candidate for the basis of acoustic-background invariant speech recognition.