Apparatus and method for sampling rate conversion

• US 5,495,432 A
• Filed: 01/03/1994
• Issued: 02/27/1996
• Est. Priority Date: 01/03/1994
• Status: Expired due to Term

• Current Assignee: Transpacific IP I Limited (Transpacific IP Group Limited)
• Original Assignee: Industrial Technology Research Institute
• Inventors: Ho, Jyh-Chern
• Primary Examiner(s): Envall, Jr., Roy N.
• Assistant Examiner(s): MOISE, EMMANUEL LIONEL
• Application Number:
  US08/177,008
• Time in Patent Office: 785 Days
• Field of Search: 364/725, 364/726, 358/12, 358/13, 358/22, 375/122, 370/70, 370/84
• US Class Current: 708/313
• CPC Class Codes: G06F 17/141 Discrete Fourier transforms
• CPC Class Codes: G06F 17/17 Function evaluation by approximation methods, e.g. inter- or extrapolation, smoothing, least mean square method G06F17/18 takes precedence ; interpolation for numerical control G05B19/18
• CPC Class Codes: H03H 17/0642 the ratio being arbitrary or irrational

Abstract

The present invention teaches a data processing system capable of performing a sampling frequency conversion. The data processing system includes a data receiving means for receiving a plurality of input data sampled at a first frequency. The data processing system further includes a discrete transforming means for performing a discrete transform on the input data at the first sampled frequency for transforming the input data to a discrete function in a second-variable-domain at the first sampling frequency. The data processing system further includes a convoluted inverse transforming means for inversely transforming the discrete function in a second-variable-domain back to the time domain at a second sampling frequency for converting the input data to a set corresponding data sampled at the second sampling frequency.

Claims

1. A data processing system capable of performing a sampling frequency conversion comprising:

a data receiving means for receiving a plurality of input data sampled at a first sampling frequency;

a discrete transforming means for performing a discrete Fourier transform (DFT) on said input data sampled at said first sampling frequency for transforming the input data to a discrete function in a second-variable-domain at said first sampling frequency by performing;

##EQU18## where N being an integer representing the number of the sampled data f(n), W₁ =e^j(2π

/N), and k=-(n-1)/2, . . . , (N-1)/2; and

a convoluted inverse transform means for performing a convolute inverse discrete Fourier transform (IDFT) transform for inversely transforming said discrete function in a second-variable-domain back to the time domain at a second sampling frequency for converting said input data to a set of corresponding data sampled at said second sampling frequency wherein said convolute inverse discrete Fourier transform (IDFT) transform being represented by;

##EQU19## where W₂ =W₁ .sup.(1/m), and m=ω

₁ /ω

₂ =f₂ /f1 which being a ratio of said second sampling frequency for said sampled data g_p to said first sampling frequency for said sampled data f(n), wherein m being any positive real number while n=0,1,2, . . . ,N-1 and where p in Equations (1-2) being a total number of the sampled data at said second sampling frequency and p={m(Tn+L)} where {K} representing an integer value of K which being smaller or equal to K and where L representing a length of a data segment which being a length of said DFT, and where T₁ =0, and

space="preserve" listing-type="equation">T.sub.n =(1/m){m(T.sub.n-1 +N)-m(T.sub.n-1 +N)} (1-3)

which being a remaining interval of a (n+1)th data segment after re-sampling at said second sampling frequency.

2. The data processing system of claim 1 wherein:

said DFT transform means further includes a first chirp transform evaluation means for evaluating said DFT for computing the values of said DFT discrete function in a second-variable-domain at said first sampling frequency; and

said convolute IDFT transform means further includes a second chirp transform evaluation means for evaluating said convoluted IDFT for computing the values of said data sampled at said second sampling frequency.

3. The data processing system of claim 2 wherein:

said DFT transform means which includes said first chirp transform evaluation means computes the values of said DFT discrete function in a second-variable-domain at said first sampling frequency by using an Equation represented by ##EQU20## where the "*" in Equation (5-1B) denotes a convolution operation and k=-(n-1)/2 . . . (n-1)/2

where F(K) corresponds to a convolution of said sampled data g_p with a sequence W₁.sup.(n*n/2) for generating an output sequence which being indexed on an independent variable K; and

said convolute IDFT transform means which includes said second chirp transform evaluation means computes the values of said data sampled at said second sampling frequency by utilizing an Equation represented as;

##EQU21## where said output sequence F(k) being re-sampled to become said sampled data g_p.

4. The data processing system of claim 3 wherein:

said DFT transform means including said first chirp transform evaluation means further includes a first multiplication means for computing the product of f(n) to W₁ -.sup.(n*n)/2 according to an Equation represented by

space="preserve" listing-type="equation">g.sub.p =f(n)W.sub.1 -.sup.(n*n)/2 ( 6-1)

said first chirp transform evaluation means further includes a first convolution means for computing the convolution e(n)*W₁ -.sup.(n*n)/2 according to an Equation represented by;

##EQU22## and said first chirp transform evaluation means further includes a second multiplication means for computing the product of said convolution e(n)*W₁ -.sup.(n*n)/2 to W₁.sup.(k*k)/2 to compute the value of F(k)according to an Equation represented by;

##EQU23## said convolute IDFT transform means including said second chirp transform evaluation means further includes a third multiplication means for computing the product of exp(jkω

₀) to W₂^k*k/2 according to an Equation represented by;

##EQU24## said second chirp transform evaluation means further includes a second convolution means for computing the convolution e(K)*W₂ -.sup.[(k*k)/2+(N-1)k/2] according to an Equation represented by;

##EQU25## and said second chirp transform evaluation means further includes a fourth multiplication means for computing the product of said convolution e(K)*W₂ -.sup.[(k*k)/2+(N-1)k/2] to W₂.sup.[-p*p/2+(n-1)**2/8] to compute the values g_p which is the data sampled at said second sampling frequency according to said Equation (5-2).

5. A data processing system capable of performing a sampling frequency conversion comprising:

a data receiving means for receiving a plurality of input data sampled at a first sampling frequency;

a discrete transforming means for performing a discrete transform on said input data sampled at said first sampling frequency for transforming the input data to a discrete function in a second-variable-domain at said first sampling frequency;

a convoluted inverse transform means for inversely transforming said discrete function in a second-variable-domain back to the time domain at a second sampling frequency for converting said input data to a set of corresponding data sampled at said second sampling frequency; and

said convoluted inverse transform means for inversely transforming said discrete function in a second-variable-domain back to the time domain at a second sampling frequency further includes a smoothing window means for applying a smoothing window to said inverse transform.

6. The data processing system of claim 5 wherein:

said smoothing window means of said convoluted inverse transform means being a symmetric low pass filter.

7. A data processing system capable of performing a sampling frequency conversion comprising:

a data receiving means for receiving a plurality of input data sampled at a first sampling frequency;

a discrete transforming means for performing a discrete transform on said input data sampled at said first sampling frequency for transforming the input data to a discrete function in a second-variable-domain at said first sampling frequency wherein said discrete transform means being a discrete Fourier transform (DFT) means;

said DFT transform means including a first chirp transform evaluation means which further includes a first multiplication means for computing the product of f(n) to W₁ -.sup.(n*n)/2 according to an Equation represented by

space="preserve" listing-type="equation">g.sub.p =f(n)W.sub.1 -.sup.(n*n)/2 ( 9-1)

said first chirp transform evaluation means further includes a first convolution means for computing the convolution e(n)*W₁ -.sup.(n*n)/2 according to an Equation represented by;

##EQU26## and said first chirp transform evaluation means further includes a second multiplication means for computing the product of said convolution e(n)*W₁ -.sup.(n*n)/2 to W₁.sup.(k*k)/2 to compute the value of F(k)according to an Equation represented by;

##EQU27## a convoluted inverse transform means for inversely transforming said discrete function in a second-variable-domain back to the time domain at a second sampling frequency for converting said input data to a set of corresponding data sampled at said second sampling frequency wherein said convolute inverse transform means being a convolute inverse discrete Fourier transform (IDFT) transform means;

said convoluted inverse transform means further including a smoothing window means which being a symmetrical low pass filter for applying a smoothing window to said inverse transform;

said convolute IDFT transform means further including a second chirp transform evaluation means which further includes a third multiplication means for computing the product of exp(jkω

₀) to W₂^k*k/2 according to an Equation represented by;

##EQU28## said second chirp transform evaluation means further includes a second convolution means for computing the convolution e(K)*W₂ -.sup.[(k*k)/2+(N-1)k/2] according to an Equation represented by;

##EQU29## and said second chirp transform evaluation means further includes a fourth multiplication means for computing the product of said convolution e(K)*W² -.sup.[(k*k)/2+(N-1)k/2] to W₂.sup.[-p*p/2+(n-1)**2/8] to compute the values g_p which is the data sampled at said second sampling frequency according to Equation 23 an Equation represented as;

##EQU30##

8. A method for sampling rate conversion comprising the steps of:

(a) receiving a plurality of input data sampled at a first sampling frequency;

(b) applying a DFT at said first sampling frequency to said input data to obtain a plurality of frequency-domain discrete functions by performing;

##EQU31## where N being an integer representing the number of the sampled data f(n), W₁ =e^j(2π

/N), and k=-(n-1)/2, . . . , (N-1)/2; and

(c) applying a convoluted IDFT to transform said frequency-domain discrete functions at said first sampling frequency to a set of corresponding time-domain data at a second sampling frequency wherein said convolute inverse discrete Fourier transform (IDFT) transform being represented by;

##EQU32## where W₂ =W₁.sup.(1/m), and m=ω

₁ /ω

₂ =f₂ /f₁ being a ratio of said second sampling frequency for said sampled data g_p to said first sampling frequency for said sampled data f(n), wherein m being any positive real number while n=0,1,2, . . . ,N-1 and where p in Equations (1-2) being a total number of the sampled data at said second sampling frequency and p={m(Tn+L)} where {K} representing an integer value of K which being smaller or equal to K and where L representing a length of a data segment which being a length of said DFT, and where T₁ =0, and

space="preserve" listing-type="equation">T.sub.n =(1/m){m(T.sub.n-1 +N)-m(T.sub.n-1 +N)} (10-3)

which being a remaining interval of a (n+1)th data segment after re-sampling at said second sampling frequency.

9. A method for sampling rate conversion comprising the steps of:

(a) receiving a plurality of input data sampled at a first frequency;

(b) performing a discrete transform on said input data at said first sampled frequency for transforming the input data to a discrete function in a second-variable-domain at said first sampling frequency wherein said discrete transform being a discrete Fourier transform (DFT) performing a first chirp transform evaluation by first computing the product of f(n) to W₁ -.sup.(n*n)/2 according to an Equation represented by;

space="preserve" listing-type="equation">g.sub.p =f(n)W.sub.1 -.sup.(n*n)/2 ( 12-1)

then computing the convolution e(n)*W₁ -.sup.(n*n)/2 according to an Equation represented by;

##EQU33## and furthermore computing the product of said convolution e(n)*W₁ -.sup.(n*n)/2 to W₁.sup.(k*k)/2 to compute the value of F(k)according to an Equation represented by;

##EQU34## (c) inversely transforming said discrete function in said second-variable-domain back to the time domain at a second sampling frequency for converting said input data to a set of corresponding data sampled at said second sampling frequency wherein said convolute inverse transform being a convolute inverse discrete Fourier transform (IDFT) transform further performing a smoothing window operation by using a symmetrical low pass filter for applying a smoothing window to said inverse transform; and

(d) said convolute IDFT transform step further applying a second chirp transform evaluation by first computing the product of exp(jkω

₀) to W₂^k*k/2 according to an Equation represented by;

##EQU35## then computing the convolution e(K)*W₂ -.sup.[(k*k)/2+(N-1)k/2] an Equation represented by;

##EQU36## and further includes a fourth multiplication means for computing the product of said convolution e(K)*W₂ -.sup.[(k*k)/2+(N-1)k/2] to W₂.sup.[-p*p/2+(n-1)**2/8] to compute the values g_p which is the data sampled at said second sampling frequency according to an Equation represented as;

##EQU37## where said output sequence F(k) being re-sampled to become said sampled data g_p.

10. A digital music synthesizer system comprising:

a data receiving and storage means for receiving and temporary storing a plurality of input data sampled at a first sampling frequency;

a pitch shifting means including a discrete transforming means for performing a discrete transform on said input data at said first sampled frequency for transforming the input data to a discrete function in a second-variable-domain at said first sampling frequency wherein said discrete transform being a discrete Fourier transform (DFT) on said input data sampled at said first sampling frequency for transforming the input data to a discrete function in a second-variable-domain at said first sampling frequency by performing;

##EQU38## where N being an integer representing the number of the sampled data f(n), W₁ =e^j(2π

/N), and k=-(n-1)/2, . . . , (N-1)/2; and

said pitch shifting means further including a convoluted inverse transform means performing a convolute inverse discrete Fourier transform (IDFT) transform for inversely transforming said discrete function in a second-variable-domain back to the time domain at a second sampling frequency for converting said input data to a set of corresponding data sampled at said second sampling frequency wherein said convolute inverse discrete Fourier transform (IDFT) transform being represented by;

##EQU39## where W₂ =W₁.sup.(1/m), and m=ω

₁ /ω

₂ =f₂ /f₁ being a ratio of said second sampling frequency for said sampled data g_p to said first sampling frequency for said sampled data f(n), wherein m being any positive real number while n=0,1,2, . . . ,N-1 and where p in Equations (1-2) being a total number of the sampled data at said second sampling frequency and p={m(Tn+L)} where {K} representing an integer value of K which being smaller or equal to K and where L representing a length of a data segment which being a length of said DFT, and where T₁ =0, and

space="preserve" listing-type="equation">T.sub.n =(1/m){m(T.sub.n-1 +N)-m(T.sub.n-1 +N)} (1-3)

which being a remaining interval of a (n+1)th data segment after re-sampling at said second sampling frequency; and

an envelop function means for performing an enveloping processing on said data at said sampling frequency whereby said data at said second sampling frequency is ready for a digital to analog conversion operation.