A Frequency Domain Technique to De-Dopplerize the Acoustic Signal from a Moving Source of Sound

Abstract

The acoustic signal emitted by a moving source of sound and sampled by a stationary microphone is distorted in amplitude and frequency by the Doppler effect. For years, recovery of the original (undistorted) source signal has been done by means of a time domain de-Dopplerization technique. This technique requires the interpolation of every data point of the microphone signal in order to remove the Doppler effect. As a consequence, this technique is computationally intensive. This paper presents an alternative de-Dopplerization technique that performs all the computations in the frequency domain. This technique requires the linearization of the relation between emission time, i.e. the time at which the signal is emitted by the source, and reception time, i.e. the time at which the signal arrives to the microphone. Then, the linear, translation, and scaling properties of the Fourier transform are used to remove the Doppler effect. In essence, this technique computes the Fourier transform of the original (undistorted) source signal directly from the Dopplerized microphone signal. Since this frequency domain de-Dopplerization technique does not require interpolation of the microphone signal, it is computationally more efficient than the traditional time domain de-Dopplerization technique.