High-accuracy, low-distortion time-frequency analysis of signals using rotated-window spectrograms
First Claim
1. A speech recognition apparatus, comprising:
- a spectral shaping source for generating a plurality of digital signal samples representative of an input speech signal;
a signal processor coupled to said source, comprising;
means for transforming said plurality of said signal samples to pre-processed signals representative of the frequency domain at various angular orientations;
means for generating initial time-frequency distributions of said pre-processed signals using analysis windows; and
means for rotating said time-frequency distributions back by said various angular orientations for generating a plurality of rotated window spectrograms, andsignal modeling apparatus for comparing the plurality of rotated window spectrograms against each of a plurality of word models and identifying the closest match.
2 Assignments
0 Petitions
Accused Products
Abstract
A speech processing and analysis apparatus and method for generating a time-frequency distribution of a speech signal combines a set of spectrograms with varying window lengths and orientations to provide a parameter-less time-frequency distribution having good joint time and frequency resolution at all angular orientations. The analysis window of a spectrogram is rotated relative to the frequency components of the signal by preprocessing using a Fractional Fourier Transform to form rotated window spectrograms. In particular, to form the rotated window spectrogram, the signal is initially pre-processed using a Fractional Fourier Transform of angle α, the spectrogram time-frequency distribution of the pre-processed signal is then computed using analysis window h(t) and then rotated by angle -α. The geometric mean of a set of rotated window spectrograms, which are indexed by both the analysis window length and the angular orientation of the window relative to the signal'"'"'s time-frequency features, is then computed to form a combination of rotated window spectrograms.
-
Citations
30 Claims
-
1. A speech recognition apparatus, comprising:
-
a spectral shaping source for generating a plurality of digital signal samples representative of an input speech signal; a signal processor coupled to said source, comprising; means for transforming said plurality of said signal samples to pre-processed signals representative of the frequency domain at various angular orientations; means for generating initial time-frequency distributions of said pre-processed signals using analysis windows; and means for rotating said time-frequency distributions back by said various angular orientations for generating a plurality of rotated window spectrograms, and signal modeling apparatus for comparing the plurality of rotated window spectrograms against each of a plurality of word models and identifying the closest match. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16)
-
-
17. A speech recognition method, comprising the steps of:
-
capturing a plurality of digital signal samples representative of an input speech signal; generating a plurality of rotated window spectrograms from said signal samples, comprising the steps of; transforming said plurality of said signal samples to pre-processed signals representative of the frequency domain at various angular orientations using Fractional Fourier Transform means; generating initial time-frequency distributions of said pre-processed signals using analysis windows; and rotating said time-frequency distributions back by said various angular orientations, and generating word models in response to said plurality of rotated window spectrograms. - View Dependent Claims (18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30)
-
Specification