Nils Westerlund

Speech enhancement for personal communication using an adaptive gain equalizer

An increasing part of our daily personal communication takes place in various noisy environments. Ever since the broad introduction of cellular phones, we tend to communicate using these phones in cars, streets and other noisy places. Noise has a negative effect on both speech intelligibility and quality; a poor signal-to-noise ratio (SNR) may indeed result in a complete lack of speech intelligibility. This paper presents a speech enhancement method for personal communication, where the input signal is divided into a number of subbands that are individually and adaptively weighted in time domain according to a short term SNR estimate in each subband at every time instant. Hence the name adaptive gain equalizer. The signal disassembly into narrow subbands is performed using computationally effective infinite impulse response filters with low group delay. The method is focused on speech enhancement, acting as a speech booster, and remains idle when the SNR in a particular subband is low. Hence, background artifacts are eliminated. In addition, the method has proven to be advantageous since it offers low complexity and low delay. It is stand-alone and works regardless of speech coding schemes and other surrounding adaptive systems.

Speech enhancement using an adaptive gain equalizer with frequency dependent parameter settings

This paper presents an enhanced noise reduction method for speech communication where the input signal is divided into a number of subbands that are individually weighted in the time domain according to the short time signal-to-noise ratio estimate (SNR) in each subband at every time instant. Instead of focusing on suppression of the noise, the method focuses on speech enhancement. The subband based method is extended with frequency dependent parameter settings, equipping the user with further tweaking possibilities. Hence, making the algorithm even more versatile and applicable in different noise situations. The method has proven to be advantageous since it offers low complexity, low delay and low distortion. Also, there is no need for a voice activity detector (VAD). The method is standalone and works regardless of speech coding schemes and other surrounding adaptive systems.

Analog circuit implementation for speech enhancement purposes

Human speech is the main method for personal communication. However, interfering noise could degrade the intelligibility of speech, eventually resulting in errors. Thus, efficient speech enhancement algorithms are needed for example in hand held battery powered hearing aids. This paper presents an implementation of a time domain method for speech enhancement purposes: the adaptive gain equalizer. The implementation is carried out on a printed circuit board using common analog electronic components, and evaluated in real-time. The proposed solution benefits from high system bandwidth, it neither quantizes nor digitalizes data, and it is likely to have more efficient power consumption as opposed to many digital signal processor (DSP) based solutions. The evaluation proves the speech enhancement performance of the analog circuit implementation.

Detection and attenuation of feedback induced howling in hearing aids using subband zero-crossing measures

A modern hearing aid should be aesthetically appealing as well as offer sufficient and adequate signal amplification. Due to the small physical size of these devices, acoustical feedback (howling) is a major problem. Apart from the annoyance and potential hearing damaging effects that howling implies, it also reduces the supplied maximum Real Ear Aided Gain (REAG). This paper proposes a novel method for subband feedback detection and cancellation, based on the zero-crossing rate measure. After splitting the hearing aid input signal into subbands, the distances between subband zero-crossings are measured. A low distance variance in a particular subband indicates that howling has arisen. The variance measure is then used to adaptively and continuously steer subband gain functions which attenuate tonal infested subbands. The method has proven to be robust and simulation indicates that it offers additional REAG of about 15 dB.

Speech enhancement by non-stationary tonal disturbance cancellation using subband zero crossing measures

A time domain based method for rejection of tonal disturbances in speech signals is presented. Tonal disturbances in personal communication arise in a variety of situations such as environmental sounds from rotating machinery, acoustic feedback in PA-systems or hearing aids and internal circuitry of some personal communication devices. The method is based on the zero crossings of subband signals. After splitting the disturbed signal into subbands, the distances between zero crossings in the subband signals are measured. A low distance spread indicates that a tonal component is dominating the subband signal. This distance spread is then used to steer a gain function that adaptively and continuously attenuates subbands in which tonal components are dominant. The method has proven to be effective on stationary as well as non-stationary disturbances with reasonable amount of speech signal distortion.