Abstract

Introduction Voice production requires the activation of a large group of muscles responsible for the laryngeal function. Laryngeal muscular activity is controlled by feedforward projections originated in motorrelated areas of the cerebral cortex, which interact with somatosensory and auditory feedbacks loops1 .Vocal compensations are evident when conversations are performed in noisy environments. In this condition, speakers involuntary increase vocal intensity and fundamental frequency, which is known as the Lombard Effect (LE) The acoustic effects due to the LE have been described extensively. However, the associated biomechanical adaptations and neural mechanisms due to the LE remain unexplored. This research describes a method to comprehensively analyze acoustic, biomechanical, and electrophysiological features of the ongoing speech production during the LE. The system allows for recordings of surface electromyography (sEMG), electroencephalography, laryngeal high-speed videoendoscopy, neck surface acceleration, electroglottograhy, oral airflow, and radiated sound pressure. Methods Pilot results considered five normal speakers that were instructed to repeatedly vocalize the phoneme /a/, and syllable /pa/ in quiet and noisy environments. Speech noise of 75 dB HL was delivered through an AD629-Interacoustics audiometer with 3045-Radioear headphones. Rate and duration of the vocalizations were controlled by presenting speech commands on a computer screen. The acoustic signal was obtained using EG2-PCX microphone. Skin vibrations were acquired by placing an accelerometer model BU-7135 Knowles at the level of the trachea. Electrophysiological signals were recorded using a BIOPAC system model-MP150. A bipolar montage was implemented for recording sEMG, in which electrodes separated at 20 mm and located 10 mm apart from the hyoid incisure were reference to the sternum. The event related potential (ERP) associated to the EEG were obtained using a bipolar montage with electrodes places on the vertex of the head (Cz) and the left mastoid. An electrode located on the midline frontal region was used as reference. Multiple correlation tests were performed to analyze the relation among the acoustic, electrophysiological and mechanical responses. Results Acoustic emissions during the LE were characterized by increased amplitude, duration and pitch. These variations were associated to compensations in the vibration pattern of the vocal folds. Likewise, they correlated the subglottic pressures estimated from the accelerometer. Furthermore, the root mean square of the sEMG elicited during the noise condition significantly differed from that obtained in quiet environments. The role of the control mechanism of the vocal production is discussed based on the variations in amp