Can Modified Casual Speech Reach The Intelligibility Of Clear Speech?

Clear speech is a speaking style adopted by speakers in an attempt to maximize the clarity of their speech and is proven to be more intelligible than casual speech. This work focuses on modifying casual speech to sound as intelligible as clear speech. First, we examine the role of speaking rate for intelligibility. Clear and casual speech signals are time-scale stretched, matching the average duration of the casual and clear speech respectively. Next, spectral shaping and dynamic range compression are considered for increasing the loudness of the original casual speech while keeping the power of signals unaffected. Subjective tests with speech-in-noise conditions using speech shaped noise at -3, 0, and 5 dB SNR show that clear speech with high speaking rate is less intelligible than the original clear speech but still more intelligible than the unmodified casual speech. However, the intelligibility score for the time-scaled modified casual speech is deteriorated. In contrast, the loudness amplification considerably improved the intelligibility of the casual speech, reaching the scores of original clear speech. Objective measurements based on Speech Intelligibility Index (SE) are well correlated with the subjective test except for the time-scaled casual signal.