Fast computation of the target signal phase in noisy environments
Introduction
Using the spatial (phase) difference between the target signal and the noise, a system can achieve the goal of noise reduction and enhance the target signal. Such a system usually consists of multi-microphones and is called beamforming system or directional system. If the noise source is moving or time varying, the adaptive directionality of such system is highly desirable so that the system could track the moving or varying noise source, otherwise, the noise reduction performance of such a system will greatly degrade. There have been a lot of available algorithms to deal with adaptive directionality so as to achieve a good compromise between the complexity and performance [1], [2], [4], [6], [7], [8]. In addition, the noise reduction performance of the beamforming system greatly depends on the number of microphones and distance of these microphones. However, in some application fields such as hearing devices, the number of microphones and distance of microphones is strictly limited. For example, behind-the-ear hearing aids can usually have only two microphones and the distance of these two microphones are about 10 mm. In these cases, most available algorithms will deliver a degraded noise reduction performance. Moreover, it would be very difficult to implement in real-time these available algorithms in these application fields mainly because of the limit of hardware size, computation speed, mismatch of microphones, power supply and other practical factors.
Based on these problems, this paper proposes an FFT based algorithm to compute the target signal phase and to implement adaptive directionality for those applications where only two microphones are available. We will show theoretically and by simulation results that the proposed scheme can reduce effectively and adaptively the noise that comes from in different direction from the one of the desired signal. We will also discuss the hardware implementation of this proposed scheme and other related practical factors. Because this proposed method needs only two FFTs and one IFFT, a simple implementation structure can be obtained and can be realizable in real-time.
The rest of this paper is organized as follows. Section 2 will present the proposed algorithm and related theoretical analysis. In Section 3, we will deal with this algorithm in more details from practical application and implementation points of view. Section 4 will present some simulation results.
Section snippets
Proposed algorithm and analysis
For a dual-microphone system, let us denote the received signals at one microphone and the other microphone as X(n) and Y(n), their DFTs as X(ω) and Y(ω), respectively. It will be proven that either of the following two processing methods can provide approximately the noise free signal under certain conditions.orwhere Z(ω) is the DFT of the system output Z(n). The conditions mainly include:
- 1.
The magnitude responses of two microphones should be
Further discussions
Concerning this proposed scheme, we will make the following discussions.
(1) From Algorithms (1), (2), we can know that this scheme can be implemented by performing two FFTs (on the basis of frame by fame) and one IFFT with the size of the frame such as 64, 128, etc. The size of the frame will be determined by the application situations. Also, for the purpose of reducing the time aliasing problem and its artifacts, windowing processing and frame overlap are required. As a matter of fact, at
Simulation results
We have simulated the proposed scheme. Four sets of simulation results are given in this paper. The first curve to the fourth curve in Fig. 1, Fig. 2 are the desired signal, the noise, the signal plus noise and the processed result corresponding to Eq. (1), respectively. The first curve to the fourth curve in Fig. 3, Fig. 4 are the desired signal, the noise, the signal plus noise and the processed result corresponding to Eq. (2), respectively. It is obvious to see that the implemented scheme
References (8)
Modified LMS algorithm for speech processing with an adaptive noise canceller
IEEE Trans. Speech Audio Process.
(1998)- et al.
An alternative approach to linearly constrained adaptive beamforming
IEEE Trans. Antennas Propagation
(1982) - et al.
Signal processing algorithms for a new, software based, digital hearing device
Hearing J.
(1998) - et al.
A comparison of hearing-aid array-processing techniques
J. Acoust. Soc. Am.
(1996)