Denoising of human speech using combined acoustic and EM sensor signal processing

Abstract

Low Power EM radar-like sensors have made it possible to measure properties of the human speech production system in real-time, without acoustic interference. This greatly enhances the quality and quantify of information for many speech related applications. See Holzrichter, Bumett, Ng, and Lea, J. Acoustic. Soc. Am. 103 ( I ) 622 (1998). By using combined Glottal-EMSensor-and Acoustic-signals, segments of voiced, unvoiced, and no-speech can be reliably defined. Real-time de-noising filters can be constructed to remove noise from the users corresponding speech signal.

Towards a Multimodal Silent Speech Interface for European Portuguese

“…1. The current approaches can be divided as follows:  Intention level (brain / Central Nerve System): Interpretation of signals from implants in the speech-motor cortex , Interpretation of signals from electro-encephalographic (EEG) sensors (Porbadnigk et al, 2009);  Articulation control (muscles): surface Electromyography (sEMG) of the articulator muscles or the larynx (Wand et al, 2009& Maier-Hein et al, 2005;  Articulation (articulators): Capture of the movement of fixed points on the articulators using Electromagnetic Articulography (EMA) sensors (Fagan et al, 2008); Real-time characterization of the vocal tract using ultra-sound (US) and optical imaging of the tongue and lips (Denby & Stone, 2004;Hueber et al, 2008); Capture movements of a talker's face through ultrasonic sensing devices (Srinivasan et al, 2010;Kalgaonkar et al 2008);  Articulation effects: Digital transformation of signals from a Non-Audible Murmur (NAM) microphone (a type of stethoscopic microphone) ) , Analysis of glottal activity using electromagnetic (Ng et al, 2000;Quatieri et al, 2006), or vibration (Patil et al, 2010) sensors; The taxonomy presented above and illustrated in Fig. 1 allows associating each type of SSI to a stage of the human speech production model providing a better understanding from where the speech information is extracted.…”

“…Automatic Speech Recognition (ASR) in the presence of environmental noise is still a hard problem to tackle in speech science (Ng et al, 2000). Another problem well described in the literature is the one concerned with elderly speech production.…”

Speech interfaces based upon surface electromyography

“…Research for extreme conditions continues to drive improvements in voice recognition accuracy and motivate approaches such as noise cancelling, bone conduction, Glottal Electromagnetic Micropower Sensors (GEMS), and visual facial analysis (Brady et al, 2004;Ng et al, 2000). Recently, attention has begun to focus on biometric sensing.…”

Small-vocabulary speech recognition using surface electromyography

“…This resulted in the development of devices such as throat microphones, interest in which continues to this day, particularly when used as part of a multi-modality speech recognition system (Graciarena et al, 2003;Shahina and Yegnanarayana, 2005;Jou et al, 2005). Military research in the area continues, focusing on sensors and techniques appropriate for communicating in noisy environments (Brady et al, 2004;Ng et al, 2000). Increasingly, researchers are experimenting with the measurement and analysis of bioelectric signals associated with speech in an effort to further minimize -or even completely eliminate -the degrading effects of acoustic noise.…”