Fusion Of Speech Spectrum Noise: Exploring The Effects Of Attention And Amplitude Modulation

Study of speech recognition mechanism in a complex acoustic environment is a very important field of hearing research, for which researchers used a variety of techniques and experimental paradigms to explore the impact of various factors of speech recognition. The process of auditory system to identify speech signals, fuse sound elements from the same sound source of speech and achieve an independent expression in compare with other sounds in the environment stimulation, has not reached an agreement. When it comes to the neural representation mechanism of speech signal, such kind of basic problems is still remaining unsolvable.The present research showed that compared to noise, speech signals has a longer echo threshold, which is significantly influenced by speech instant fluctuations. In terms of other auditory researches, the processing of amplitude characteristics may also be affected by top-down advanced cognitive process, like attention. The present research added attention conditions into the investigation of the neural characterization on speech acoustic amplitude information such as frequency. First, we found speech spectrum noise sounds with different amplitude modulation frequencies induced relevant transient as well as steady-state responses, then made comparison between these results with same stimulations without attention in order to find out whether there is interference between amplitude modulation and attention.This study used event-related potentials and eeglab steady-state response analysis technology to investigate two experiments, attended and unattended conditions, neural responses of different modulation frequency stimuli. These experiments applied the mixed design with two factors. The first factor contained 6 amplitude modulation frequencies conditions, and the other factor contained two attention states.The experimental results showed that the attention conditions induced by different modulation frequency of the sound were significantly different both in ERPs and steady state responses. Compared to the steady-state spectral noise, 2Hz, 4Hz,10 Hz modulation frequency steady-state speech spectrum noise induced higher P2 components peak. It was found that the 2Hz and 4Hz sound stimuli evoked a significant steady-state response at the corresponding low frequency area in the EEG domain of time-frequency energy spectrum. In the unattended condition, the N1 and P2 components showed no significant difference, and the energy distribution of the steady-state response was rather dispersed. Mixed design analysis of ANOVA implied that different sound stimulis’ main effect was significant, that of attention condition was remarkable, and the interaction wasn’t significant at all. In conclusion of the two experimental ideas, it is clearly and remarkably proved that the auditory cortex has a selective encoding preference for the low AMf feature, which is regulated by the attention.