Patent
Digital filtering method, digital filtering equipment, digital filtering program, and recording medium and recorded device which are readable on computer
Reads0
Chats0
TLDR
In this article, the feature quantity can be extracted only by applying an integer summation and comparison operation, so that performance of arithmetic operation is extremely simplified to be fast and inexpensive.Abstract:
Included in the digital filtering equipment for extracting a feature quantity from a speech signal in order to execute a speech recognition based on an inputted speech signal are: a) an waveform determining section for obtaining an inputted speech signal and quantizing the speech signal waveform; b) a division value operating section for summing a quantized signal data in a prescribed adjoining region to divide the summation value by the number of summed data, with respect to a data quantized at each point by the waveform determining section, whereby a division value is obtained with the data being centered; c) a comparison section for comparing a division value calculated by the division value operating section and the quantized data calculated in the division value operating section to output logical truth of a comparison result, with respect to each data; and d) a conversion section for converting the quantized data into the selected data points based on an output from the comparison section. Thus, the feature quantity can be extracted only by applying an integer summation and comparison operation, so that performance of arithmetic operation is extremely simplified to be fast and inexpensive.read more
Citations
More filters
Patent
Apparatus and Method for Processing an Audio Signal for Speech Enhancement Using a Feature Extraction
TL;DR: In this article, a feature extractor is used to extract at least one feature per frequency band of a plurality of frequency bands of a short-time spectral representation, where the feature represents a spectral shape of the short time spectral representation in the frequency band.
Patent
Dynamically adding or removing functionality to speech recognition systems
TL;DR: In this paper, a system and method of changing features of an existing automatic speech recognition (ASR) system includes monitoring speech received from a vehicle occupant for one or more keywords identifying a feature to remove from or add to the ASR system.
Patent
Distributed audio coding for wireless hearing aids
Olivier Roy,Martin Vetterli +1 more
TL;DR: In this paper, a method for computing ICLD from a first and second audio source signals, the first source signal being wired with a first processing module and the second source signal with a second processing module, was presented.
Patent
Noise detection apparatus, noise removal apparatus, and noise detection method
TL;DR: In this article, a time-frequency transform unit is used to transform an input signal from a time domain to a frequency domain to produce a spectrum, a power spectrum calculating unit is configured to obtain powers of frequencies from the spectrum, and a peak stationarity detecting unit is defined to detect frequencies at which a stationary peak of the powers exists.
PatentDOI
Vocal fry detecting apparatus
TL;DR: In this article, a very short-term peak detection processing unit was used to detect vocal fry (VF) in a speech signal with a first frame length and first frame shift amount and detecting each power peak.
References
More filters
Patent
Lossless coding method for waveform data
TL;DR: In this paper, a quantiser is jointly responsive to a first sample value of a signal input to the prediction filter and an output value of the quantiser at a previous sample incident.
Patent
Digital speech detector
TL;DR: In this paper, the authors improved the detection sensitivity and noise rejection of an arrangement for detecting speech in the presence of noise by accumulating the weighted differences between input signal samples and their short-term running average.
Patent
Speech processing method and apparatus and program therefor
TL;DR: In this paper, a scheme to judge emphasized speech portions is proposed, wherein the judgment is executed by a statistical processing in terms of a set of speech parameters including a fundamental frequency, power and a temporal variation of a dynamic measure and/or their derivatives.
Patent
Method and device for voice detection and its recording medium
Atsushi Murashima,淳 村島 +1 more
TL;DR: In this article, a first filter (2061 of Figure 1) computes a long time average of first fluctuation amount based on the difference between the line spctrum frequency of inputted voice signals and its long-time average.
PatentDOI
Echo canceller ensuring further reduction in residual echo
TL;DR: In this paper, an echo cancellation part canceling echo included in a near-end input signal, a noise suppression part employing a low-pass filter having a variable cut-off frequency adaptive to power of an input signal and a determination part making a determination based on a comparison between a power level of the output of the cancellation part and the variable threshold.