【the Invention】 [discrete Wavelet Transform Based Multiple Template-matching for Speech Recognition]
نویسنده
چکیده
【Abstract】 This invention is method of speech recognition that consists of plural overlapped templatematching (TM). In order to economize calculations of TM, the data of low resolution given by Haar discrete wavelet transform (HDWT) is used. Templates of wavelet coefficient (WC) on waveform are used for recognition of phoneme. A sum of WC in a scale (SWC) corresponds to a frequency component on the waveform. The templates of SWC on a waveform for short syllable are used for recognition of the syllable. [Recognition on phoneme] Three kinds of frame i.e. 64 of data (6.4msec), 128 (12.8msec) and 256 (25.6msec) were picked up from waveform where each segment begins at each peak of wave, and each segment of wave is transformed wavelet coefficients (WC) and the number of WC can be decreased up to 15 by the multiple resolution characteristics of WT. A sum of absolute value of WC in each scale (SWC) indicates frequency components on the scale. That is, frequency distribution obtained from distribution of the SWC. The principal constituents of voices of a male are observed at scale on 6.4msec (156Hz), 3.2msec (312.5Hz), 1.6msec (625Hz), 0.8msec (1.25 kHz), and 0.4msec (2.5 kHz). Since the ratio of SWC (the 3.2msec band)/SWC (the 1.6msec band) and SWC (the 6.4msec band)/SWC (the 1.6msec band) became a node (value=1) at the transition region of phoneme, segmentation of phoneme in continuous speech is able to find out. [Recognition of syllable] As for recognition of syllable, data size for a frame increases. Data on 204.8msec for a frame are shifted every 6.4msec were transferred to a set of WC and sum of absolute value of WC in each scale (SWC). (SWC) is used as a feature vector of TM on syllable. [Overlapped TM for recognition of voice] Overlapped TM is used for an integration of results. Any kind of TM is possible in a brain. A block diagram of layered TM system for presenting voice recognizer is shown in Fig.1. Fig.1 A speech recognition as an application of multiple-template matching
منابع مشابه
Speech Recognisation System Using Wavelet Transform
To develop speech recognition system with low word error rate using wavelet transform through pattern recognition approach. The aim of this paper is to make intelligent system that can recognize the speech signal. This includes also how the feature extracted from the speech signal using Discrete Wavelet Transform and then Dynamic Time Warping is used for pattern matching from the stored databas...
متن کاملPattern recognition technique for integrated circuit (IC) pins inspection using wavelet transform with chain- code-discrete fourier transform and signal correlation
This paper proposes a new technique for recognizing integrated circuit (IC) pins in an IC chip. The proposed technique applies the wavelet transform and discrete Fourier’s transform to extract the interesting features for classifying the IC pin pattern. The accuracy and inspection time of the proposed method are investigated by comparing with other pattern recognition techniques such as full te...
متن کاملA comparison of performance monitoring approaches to fusing spectrogram channels in speech recognition
Implementations of two performance monitoring approaches to feature channel integration in robust automatic speech recognition are presented. These approaches combine multiple feature channels, where the first one uses a feed-forward entropy-based criterion and the second one, motivated by psychophysical evidence in human speech perception, employs a closed loop criterion relating to the overal...
متن کاملImage Based Spam Email Detection by Combined Approach
Electronic mail is one of the important communication channels of information technology which serves across the globe. Though the functionalities of e-mail have been very helpful in serving both individuals and institutions, it encounters a major issue called ‘Spamming’. Spam mails are unwanted text or image-based messages, often sent without the consent of users so as to fill their mailboxes....
متن کاملAutomatic target shape recognition via deformable wavelet templates
In this research, we propose the deformable wavelet template (DWT) for object shape description. DWT o ers not only the global information at the lower scales, but also local features at higher di erential scales. Thus, it provides a natural tool for multiresolution representation and can be used conveniently in a hierarchical matching procedure. In the presentation, we address the three main p...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2007