Frequency analysis method

• US 5,583,784 A
• Filed: 05/12/1994
• Issued: 12/10/1996
• Est. Priority Date: 05/14/1993
• Status: Expired due to Term

• Current Assignee: Fraunhofer Gesellschaft Zur Foerderung Der Angewandten Forsching E.V.
• Original Assignee: Fraunhofer Gesellschaft Zur Foerderung Der Angewandten Forsching E.V.
• Inventors: Kapust, Rolf, Seltzer, Dieter
• Primary Examiner(s): Voeltz, Emanuel T.
• Assistant Examiner(s): SMITH-STEWART, DEMETRA R
• Application Number:
  US08/241,851
• Time in Patent Office: 943 Days
• Field of Search: 381/36, 381/31, 381/35, 381/36, 381/37, 395/2
• US Class Current: 702/77
• CPC Class Codes: G10L 25/27 characterised by the analysis technique
• CPC Class Codes: G10L 25/48 specially adapted for particular use
• CPC Class Codes: H04B 1/665 using psychoacoustic properties of the ear, e.g. masking effect

Abstract

A frequency analysis method comprises using a window function to evaluate aemporal input signal present in the form of discrete sampled values. The windowed input signal is subsequently subjected to Fourier transformation for the purpose of generating a set of coefficients. In order to develop such a method so that the characteristics of the human ear are simulated not only with respect to the spectral projection in the frequency range, but also with respect to the resolution in the temporal range, a set of different window functions is used to evaluate a block of the input signal in order to generate a set of blocks, weighted with the respective window functions, of sampled values whose Fourier transforms have different bandwidths, before each of the simultaneously generated blocks of sampled values is subjected to a dedicated Fourier transformation in such a way that for each window function at least respectively one coefficient is calculated which is assigned the bandwidth of the Fourier transforms of this window function, and that the coefficients are chosen such that the frequency bands assigned to them essentially adjoin one another.

Claims

1. A frequency analysis method that simulates perception characteristics of human hearing, comprising the steps of:

extracting from an input signal a sequence of blocks of discrete sampled values, wherein successive blocks of the sequence overlap each other in time;

concurrently windowing one of the successive blocks using a plurality of differing, symmetrical window functions of substantially identical length, each window function having a corresponding Fourier transform with a different bandwidth, to produce a plurality of windowed blocks of weighted sampled values; and

performing a Fourier transformation on each windowed block wherein for each windowed block at least one coefficient is generated having assigned thereto the bandwidth of the Fourier transform of the window function used to produce the windowed block, and wherein the at least one coefficient for each windowed block is chosen such that bandwidths assigned to the chosen coefficients of the plurality of windowed blocks adjoin one another approximately without gaps and in a manner approximately free from overlapping.

2. The frequency analysis method according to claim 1, wherein the step of windowing further includes a step of:

weighting each of the plurality of windowed blocks by applying differing sinx/x window functions having zero crossings situated at sampling instants of the discrete sampled values, a temporal resolution of the windowed block corresponding to a mutual spacing of the zero crossings, and the temporal resolution being inversely proportional to twice the bandwidth of the Fourier transform of the window function to which the Fourier transform corresponds.

3. The frequency analysis method according to claim 2, wherein the step of performing the Fourier transformation includes steps of:

performing a discrete fast Fourier transformation (FFT) which delivers in a frequency domain a number of coefficients corresponding to a number of input values in a time domain of a respective window function, having a separation in the frequency domain between each of the number of coefficients that corresponds to a sampling frequency of the input signal divided by a length of the discrete fast Fourier transformation, and whose bandwidth is prescribed by the bandwidth of the Fourier transform of the respective window function; and

selecting only those coefficients from the number of coefficients by which it is possible, in conjunction with coefficients available in a time-parallel fashion from discrete fast Fourier transformations on others of the plurality of windowed blocks of weighted sampled values, to construct a cascade of band filters in a case of other centre frequencies, so that resulting frequency bands adjoin one another approximately without gaps.

4. The frequency analysis method according to claim 2, wherein

to improve selection properties of a cascade of band filters produced by the method, a length of the plurality of differing window functions is a multiple of a Fourier transformation length of the corresponding Fourier transform, and wherein the method further comprises a step of,

before transformation, reducing a multiplicity of windowed discrete sampled values to the Fourier transformation length by division and additive superposition of the windowed discrete sampled values.

5. The frequency analysis method according claim 2, wherein

a first of the at least one chosen coefficient is generated from the input signal using a second order Goertzel algorithm.

6. The frequency analysis method according to one of claims 1 to 5, further comprising:

an iterative method for determining discrete centre frequencies and band limits for defining the window functions, the iterative method including steps of;

(A) setting a running frequency variable (f_centre) to a start frequency (f_top) which lies on a frequency raster chosen by a length of the Fourier transformation;

(B) displacing the running frequency variable (f_centre) by one step (df) on the frequency raster;

(C) determining a bandwidth resulting from a given, nonlinear function of the centre frequency; and

(D) repeating steps B and C until a band limit of the band belonging to the running frequency variable (f_centre) coincides with a band limit of a previously determined band.

7. The frequency analysis method according to claim 6, wherein

the given, nonlinear function is a relationship between the centre frequency and a frequency bandwidth in which acoustic events are jointly perceived.