Kimitoshi Fukudome

Automatic Cancellation Method of Multipath Echo Distortion in FM Broadcasting Receivers

An automatic canceling method of multipath echo distortion in FM broadcasting receiver is proposed. The cancelling system is composed of two additional parts to the conventional receiver: a programable transversal filter operating at linear IF stage and a microcomputer. The microcomputer calculates echo parameters (relative echo amplitudes, delay times and phase differences between direct wave and echoes at the carrier frequency), from the amplitude-frequency characteristics between sending station and IF stage of the receiver, which is obtained from both outputs of an IF signal envelope detector and FM demodulator. The microcomputer sends control signals to the weighting circuits of the transversal filter. The output signal of the transversal filter is fed back and added to the incoming signal. Tap weighting adjustment is continued until a flat amplitude-frequency characteristics is obtained. The results of computer simulation show this method works well for various echo conditions and the distortion can be completely eliminated.

Observation of reverberation curves by two relatively prime pseudorandom sequences

The pseudorandom sequence whose periods are relatively prime are introduced to measure reverberation decay curves in a room. One pseudorandom sequence is driven by d-times faster clock than the other and is switched after the polarity of the other sequence. After radiating the switched signal from a loudspeaker, the crosscorrelation between the switching sequence and d-times decimated envelope sequence of a microphone output is calculated. The decay curve obtained by the integration of the crosscorrelation through the formula given by Schroeder equals the ensemble average of the envelope of decay curves driven by white noise. It is shown that the decimation serves not only to decrease the amount of computation but also to increase the signal to noise ratio of the measured decay curves.

Intelligibility improvement of cellular phone speech in a noisy environment

In a noisy environment cellular phone users often have difficulties in hearing far‐end speech because the ambient noise disturbs the user’s understanding of the conversation. To improve speech intelligibility in noise, speech enhancement is a feasible solution. In this paper the effects of two speech enhancement methods, multiband amplitude compression and formant enhancement, are reported. The multiband amplitude compression is a typical method for hearing aids that divides speech signal into some frequency bands and conducts amplitude compression independently. On the other hand, the formant enhancement is an LPC analysis‐synthesis technique that provides for formant frequency shaping. The performances of these methods on the speech intelligibility were evaluated in the simulated real‐world noise circumstances. Recorded noise sounds were presented from loudspeakers in a soundproof room and the speech samples were reproduced from real cellular phone handset which was fixed on the subject’s head. VCV (vowel–consonant–vowel sequence) units were used as test speech samples. The results show that multiband amplitude compression makes significant improvement in speech intelligibility and does not degrade the subjective impression of speech.

A method of interpolating the diffractive information of the sphere‐baffled microphone in the sound field of the spherical wave

The sphere‐baffled microphone (SBM) has been used for estimating DOA of progressive plane waves and their extractions, but it is now planned to construct the diffractive information directory (DID) of the SBM with DOA entry and source—distance entry in the sound field of spherical wave. This paper deals with the interpolation methods of getting information associated with entries not appearing in DID. Numerical computation of the diffractive information is done for the 2‐element SBM (right and left microphones at both ends of a diameter of 17.68 cm) at frequency points corresponding to the sampling frequency 22.05 kHz and 1024 point DFT. Two data sets of DOA interval of 6° [0,6,12,...] and [3,9,15,...] are used for interpolation and evaluation of errors. Interpolations and their error evaluations are done in the regions of both the log magnitude versus frequency and the group delay versus frequency. Both the bilinear interpolation and cubic interpolation method result in significant interpolation errors u...

Superdirectivity design for a sphere‐buffled microphone array

For the signal extraction from a diffuse sound field, the superdirecitonal microphone array seems to be more stable and more effective than the noise canceling microphone array system. In this research, for the sphere‐baffled microphone array (SBM), the superdirectivity characteristic was designed by the frequency domain beamforming method and the complex least mean square algorithm. The frequency domain beamforming is a method for a broadband beamforming. There is a possibility of improvement of the directivity frequency characteristic. The SBM is an array which was embedded 36 elements on the rigid diffractive sphere of 8.84 cm radius. An advantage of SBM is its isotropy. When the directivity characteristic was fixed for a definite direction, only the shift coefficent can be used to change the mainlobe direction. This idea can be easily generalized into the three‐dimensional mainlobe control. In a typical design calculation, for the beamwidth of −3 dB, 30° was achieved at the frequency 2 kHz. This characteristic is comparable to that of a larger size microphone, which is called a shotgun microphone.

Three‐dimensional extraction and reproduction of signals from an interfering sound field

This paper describes a method for the three‐dimensional reproduction of any constituent sound signals existing in the interfering sound field. The method is composed of three parts. (1) Estimation of incident directions and spectra of unknown sound sources using sphere‐baffled microphones and the diffractive information of the sphere [K. Fukudome, J. Acoust. Soc. Jpn. 44, 272–281 (1988)]. (2) Extraction of sound source signals by adding the output signals of the extracting filters that are fed by the sphere‐baffled microphones. Characteristics of the filters are determined according to the estimated results of the incident directions and the diffractive information. (3) Three‐dimensional reproduction of any constituent sound signals using headphones. The right and left signals of the headphones are generated after convolving the head‐related transfer characteristics with the constituent sound source signal. The validity of the present method has been verified by computer simulation. Finally, the problems ...

Observation impulse response by two relatively prime pseudorandom sequences

A crosscorrelation method using two pseudo-random sequences whose periods are relatively prime is introduced to measure the impulse response of a bandpass system with improved signal-to-noise ratio. One pseudorandom sequence is driven by d-times faster clock than the other and is switched after the polarity of the other sequence. The switched signal is used as an input of the bandpass system. The PR-sequence, which is defined as an arithmetic product of the slower sequence and the decimated faster sequence by d, is prepared. It is shown that the cross-correlation between the PR-sequence and the decimated output of the system equals the impulse response of the system. The example of the practical implementation of this system is described for room acoustics measurement. The experimental results are shown and the final signal-to-noise ratio is compared with the theoretial values.

Signal processing technique for reducing tracing distortion in two‐dimensional recording systems

Because of high density recording and very short access time (less than 0.1 s) for 104−105 audio program sources of duration 10 s or so, a microfiche system which records audio signals as two‐dimensional patterns after suitable signal processing is now being developed. Simulations are carried out on several candidates of signal processing methods in order to overcome serious distortion produced by mistracing of the two‐dimensionally transformed audio patterns. Among them, recoding of pitch or noise and impulse response of vocal tract extracted by PARCOR method is wonderfully excellent and allows oblique mistracing over up to 16 lines of recorded two‐dimensional signals without distortion for speech. Studies on general audio signals other than speech are continuing.