Daniel Willem Elisabeth Schobben

Acoustic ear recognition for person identification

In this paper we investigate how the acoustic properties of the pinna-i.e., the outer flap of the ear- and the ear canal can be used as a biometric. The acoustic properties can be measured relatively easy with an inexpensive sensor and feature vectors can be derived with little effort. We achieve equal error rates in the order of 1.5%-7%, depending on the application device that is used to do the measurement.

Privacy protected biometric templates: acoustic ear identification

Unique Biometric Identifiers offer a very convenient way for human identification and authentication. In contrast to passwords they have hence the advantage that they can not be forgotten or lost. In order to set-up a biometric identification/authentication system, reference data have to be stored in a central database. As biometric identifiers are unique for a human being, the derived templates comprise unique, sensitive and therefore private information about a person. This is why many people are reluctant to accept a system based on biometric identification. Consequently, the stored templates have to be handled with care and protected against misuse [1, 2, 3, 4, 5, 6]. It is clear that techniques from cryptography can be used to achieve privacy. However, as biometric data are noisy, and cryptographic functions are by construction very sensitive to small changes in their input, and hence one can not apply those crypto techniques straightforwardly. In this paper we show the feasibility of the techniques developed in [5], [6] by applying them to experimental biometric data. As biometric identifier we have choosen the shape of the inner ear-canal, which is obtained by measuring the headphone-to-ear-canal Transfer Functions (HpTFs) which are known to be person dependent [7].

Real-time Adaptive Concepts in Acoustics: Blind Signal Separation and Multichannel Echo Cancellation

List of Figures. Preface. Part I: Background and Introduction. 1. Introduction. 2. Array Processing Techniques. 3. Efficient Filtering Using FFTS. Part II: Acoustic Echo Cancellation. 4. An Efficient Adaptive Filter Implementation. 5. Efficient Multichannel RLS. Part III: Blind Signal Separation. 6. Blind Signal Separation, An Overview. 7. A Blind Signal Separation Algorithm. 8. Joint Blind Signal Separation and Echo Cancellation. 9. Blind Signal Separation Algorithm Evaluation. 10. Conclusions. Appendices.

Acoustic ear recognition

We investigate how the acoustic properties of the pinna – i.e. the outer flap of the ear- and the ear canal can be used as a biometric. The acoustic properties can be measured relatively easy with an inexpensive sensor and feature vectors can be derived with little effort. Classification results for three platforms are given (headphone, earphone, mobile phone) using noise as an input signal. Furthermore, preliminary results are given for the mobile phone platform where we use music as an input signal. We achieve equal error rates in the order of 1%-5%, depending on the platform that is used to do the measurement.

Blind Signal Separation Algorithm Evaluation

Recently, many new Blind Signal Separation (BSS) algorithms have been introduced. Authors evaluate the performance of their algorithms in various ways. Among these are speech recognition rates, plots of separated signals, plots of cascaded mixinglunmixing impulse responses and signal to noise ratios. Clearly, not all of these methods give a good reflection of the performance of these algorithms. Moreover, since the evaluation is done using different measures and different data, results cannot be compared. As a solution a unified methodology of evaluating BSS algorithms is provided along with data that is available such that researches can compare their results. Although this chapter is focussed on acoustic applications, many of the remarks apply to other BSS application areas as well. The work in this chapter is published in Schobben eta!.,19991.

Efficient Multichannel RLS

This chapter presents a computationally efficient adaptive multichannel algorithm which is derived from the well-known multichannel recursive least squares algorithm. The new algorithm is a generalization of the block frequency domain adaptive filter which is a computationally efficient approximation of the classical recursive least squares algorithm. Application areas in the audio context include stereo echo cancellation for teleconferencing, hands-free telephony and voice-controlled machinery. In such applications, both the desired signal and acoustic echoes that are coming from multiple loudspeakers are picked up by a microphone. Applying multiple single-channel adaptive filters to eliminate such echoes leads to slow convergence when the loudspeaker signals are mutually correlated. This chapter presents a multichannel adaptive filter which uses a joint update rule for all filters. Experiments are carried out to confirm the operation of the multichannel adaptive filter and compare it with the performance of separately updated adaptive filters.

Efficient Filtering Using FFTS

This chapter gives an introduction to the efficient implementation of filters in the frequency domain. A background of design aspects is given to facilitate for the real-time implementation of such filters for a given latency (i.e. the delay between the input and the output of a filter). This chapter may be skipped when the reader is familiar with the overlap-save technique in the frequency domain and has no interest in new insights on low latency frequency domain filters.

A Blind Signal Separation Algorithm

This chapter addresses the problem of separating multiple speakers from mixtures of these that are obtained using multiple microphones in a room. A new blind signal separation algorithm is derived which is entirely based on second order statistics. The algorithm can run in off-line or online (adaptive) mode. One of the advantages of this algorithm is that no assumptions are made about the probability density functions or other properties of the signals. The blind signal separation algorithm has been tested using microphones that pickup speech signals of speakers that are talking simultaneously to give an indication of its performance for real-world data.

Joint Blind Signal Separation And Echo Cancellation

The problem of joint blind signal separation and acoustic echo cancellation arises in applications such as teleconferencing and voice-controlled machinery. Typically, in the same room as the local speakers there are loudspeakers reproducing far-end speech or music signals. The contributions of these loudspeaker signals to the microphone signals need to be canceled. The remaining signals are then separated so that the individual local speakers are recovered. In the previous chapter the CoBliSS blind signal separation algorithm is introduced. In this chapter an extension of CoBliSS is presented (ECoBliSS) which can simultaneously deal with blind signal separation and echo cancellation. The computational complexity of ECoBliSS is less than that of CoBliSS when processing the same number of signals. The performance of the extended CoBliSS algorithm is evaluated using audio recorded in a real acoustic environment. This chapter is based on Schobben and Sommen, 1999a.

Blind Signal Separation, An Overview

Blind Signal Separation (BSS) is the process of recovering independent signals that correspond to the individual source signals using only observed linear mixtures of these. In an acoustic context, these source signals are correlated in time and are assumed to be independent of each other. The mixing system is convolutive in the sense that the microphones pick up delayed and attenuated versions of the source signals due to reflections in the room. The microphone signals typically contain some microphone noise.