Signal processing apparatus

• US 10,339,907 B2
• Filed: 01/22/2018
• Issued: 07/02/2019
• Est. Priority Date: 03/15/2017
• Status: Active Grant

• Current Assignee: Casio Computer Company Limited
• Original Assignee: Casio Computer Company Limited
• Inventors: Yokota, Masuo
• Primary Examiner(s): Fletcher, Marlon T
• Application Number:
  US15/876,406
• Publication Number:
  US 20180268794A1
• Time in Patent Office: 526 Days
• CPC Class Codes: G10H 1/043 Continuous modulation
• CPC Class Codes: G10H 1/125 using a digital filter
• CPC Class Codes: G10H 2210/281 Reverberation or echo
• CPC Class Codes: G10H 2250/115 FIR impulse, e.g. for echoes or room acoustics, the shape of the impulse response is specified in particular according to delay times FIR filters for musical processing G10H2250/075
• CPC Class Codes: G10H 2250/145 Convolution, e.g. of a music input signal with a desired impulse response to compute an output transforms, i.e. mathematical transforms into domains appropriate for musical signal processing, coding or compression G10H2250/215
• CPC Class Codes: H03H 17/0213 Frequency domain filters using Fourier transforms
• CPC Class Codes: H03H 17/06 Non-recursive filters

Abstract

A signal processing apparatus has a first memory in which plural pieces of FIR coefficient data used for implementing an FIR filter algorithm are stored, a second memory which stores plural pieces of input data to be subjected to the FIR filter algorithm, and a processor implements the FIR filter algorithm using the plural pieces of FIR coefficient data stored in the first memory and the plural pieces of input data stored in the second memory as many times as the number corresponding to a designated filter order, in which filter algorithm each piece of coefficient data and each piece of input data are multiplied together and resultant products are summed up. The signal processing apparatus is provided, which can implement plural sorts of FIR filter algorithms of filter order which can be changed flexibly.

Claims

1. A signal processing apparatus comprising:

a first memory which stores plural pieces of coefficient data used for implementing a filter algorithm and containing plural coefficient data sets as many as a number corresponding to a filter order of the filter algorithm for at least one sort of filter algorithm of a variable filter order;

a second memory which stores input data sets from a last sample to a sample before the (filter order-1)-th sample for every filter algorithm as delay data sets; and

a processor which receives plural pieces of input data to be subjected to the filter algorithm, executes the filter algorithm using the plural pieces of coefficient data stored in the first memory and the plural pieces of received input data as many times as a number corresponding to a designated filter order, and outputs an execution result of the filter algorithm, in which filter algorithm each piece of coefficient data and each piece of input data are multiplied together and resultant products are summed up, and further, successively receives the coefficient data stored in the first memory and current input data stored in the second memory, repeatedly implements a first-order filter algorithm by time division processing of a dividing time length corresponding to the filter order, and outputs output-data within each sampling period.

2. The signal processing apparatus according to claim 1, wherein:

the second memory stores the plural pieces of received input data;

the filter algorithm is a Finite Impulse Response (FIR) filter algorithm;

the first memory stores plural pieces of FIR coefficient data; and

the processor implements the FIR filter algorithm using the plural pieces of FIR coefficient data stored in the first memory and the plural pieces of received input data stored in the second memory as many times as the number corresponding to a designated filter order, in which filter algorithm each piece of FIR coefficient data and each piece of input data are multiplied together and resultant products are summed up.

3. The signal processing apparatus according to claim 1, wherein:

the processor is configured to designate a number corresponding to the filter order of the filter algorithm to be implemented within a predetermined number;

the first memory stores at least the predetermined number of pieces of coefficient data; and

the second memory stores at least the predetermined number of pieces of input data.

4. The signal processing apparatus according to claim 1, wherein the processor:

assigns continuous time durations of a time length counted by a clock having a cycle subdividing a sampling period to the filter algorithms respectively, each time length being long enough for implementing the filter algorithm of the filter order in the sampling period,

successively supplies the coefficient data stored in the first memory and the current input data or the delay data stored in the second memory to the filter algorithm implementing unit, and makes the filter algorithm implementing unit implement the filter algorithm in the continuous time duration, in which filter algorithm the successively supplied data are multiplied together and resultant products are summed up in synchronism with the clock, and

controls the filter algorithm implementing unit to output output-data of the filter algorithms in each sampling period.

5. The signal processing apparatus according to claim 1, further comprising a filter algorithm implementing unit, the filter algorithm implementing unit comprising:

a first register which retains in synchronism with the clock the coefficient data output from the first memory;

a first selector which selects either one of input data in the current sampling period and the delay data output from the second memory;

a second register which retains in synchronism with the clock the data output from the first selector;

a multiplier which multiplies the coefficient data output from the first register and the data output from the second register to obtain product data;

a third register which retains in synchronism with the clock the product data output from the multiplier;

an adder which adds the product data output from the third register and current summed-up data together to obtain sum data;

a fourth register which retains in synchronism with the clock the sum data output from the adder; and

a second selector which selects either one of data of a value “

0” and

the sum data output from the fourth register to feed back the selected data as the summed-up data to the adder.

6. An effect applying apparatus comprising:

a first memory which stores plural pieces of Finite Impulse Response (FIR) coefficient data used for implementing an FIR filter algorithm; and

a processor which receives plural pieces of input data to be subjected to the FIR filter algorithm, the plural pieces of received input data being stored in a second memory, uses the plural pieces of FIR coefficient data stored in the first memory and the plural pieces of received input data stored in the second memory to multiply each piece of FIR coefficient data and each piece of input data together as many times as a number corresponding to a designated filter order, and sums up resultant products,

wherein, when an acoustic effect is applied to an original tone by convoluting the original tone with impulse response data of an acoustic effect tone, the processor;

implements the FIR filter algorithm to convolute the original tone with first half data of the impulse response data in the time domain of a unit of a sampling period; and

further implements a Fast Fourier Transform (FFT) algorithm on second half data of the impulse response data in the frequency domain of a unit of a block, thereby convoluting the original tone with the second half data.

7. The effect applying apparatus according to claim 6, wherein the first half data and the second half data of the impulse response data are set arbitrarily.

8. The effect applying apparatus according to claim 6, wherein the processor decides a block size of the block and sets a filter order of the FIR filter algorithm and an FFT point number of the FFT algorithm to twice as large as the block size.

9. The effect applying apparatus according to claim 8, wherein the processor decides the block size in accordance with a time length of the impulse response data.

10. The effect applying apparatus according to claim 8, wherein the processor decides the block size in accordance with a tone generation setting condition of the original tone.

11. The effect applying apparatus according to claim 8, wherein the processor:

stores coefficient data in the time domain of the impulse response data;

after deciding the block size, removes FIR coefficient data of a filter order equivalent to twice as large as the block size to be used for implementing the FIR filter algorithm from the coefficient data of the impulse response data to obtain remaining coefficient data; and

implements the FFT algorithm on the remaining coefficient data to transform the same data to coefficient data for convolution in the frequency domain, expanding the same on a memory.

12. The effect applying apparatus according to claim 11, wherein the processor:

adds data indicating the decided block size to the coefficient data for convolution in the frequency domain and stores resultant data; and

when the block size is decided again and only when the decided block size is different from the previously stored block size, implements the FFT algorithm on the coefficient data for convolution in the frequency domain, thereby renewing the same coefficient data.

13. A method for implementing a filter algorithm under control of a processor of a signal processing apparatus, the signal processing apparatus comprising a first memory which stores plural pieces of coefficient data to be used for implementing the filter algorithm and containing plural coefficient data sets as many as a number corresponding to a filter order of the filter algorithm for at least one sort of filter algorithm of a variable filter order, and a second memory which stores input data sets from a last sample to a sample before the (filter-order-1)-th sample for every filter algorithm as delay data sets, the method comprising:

receiving plural pieces of input data to be subjected to the filter algorithm;

designating a filter order of the filter algorithm;

implementing the filter algorithm using the plural pieces of coefficient data storing unit stored in the first memory and the plural pieces of received input data as many times as a number corresponding to the designated filter order, in which filter algorithm each piece of coefficient data and each piece of input data are multiplied together and resultant products are summed up, and outputting a result of implementing the filter algorithm,

wherein the implementing the filter algorithm comprises successively receiving the coefficient data stored in the first memory and current input data stored in the second memory, repeatedly implementing a first-order filter algorithm by time division processing of a dividing time length corresponding to the filter order, and outputting output-data within each sampling period.

14. A method for implementing a Finite Impulse Response (FIR) filter algorithm under control of a processor of an effect applying apparatus, the effect applying apparatus comprising a first memory which stores plural pieces of Finite Impulse Response (FIR) coefficient data used for implementing the FIR filter algorithm, and a second memory, and the method comprising:

storing plural pieces of received input data to be subjected to the FIR filter algorithm in the second memory; and

implementing the FIR filter algorithm using the plural pieces of FIR coefficient data stored in the first memory and the plural pieces of received input data stored in the second memory as many times as a number corresponding to a designated filter order, in which filter algorithm each piece of FIR coefficient data and each piece of input data are multiplied together and resultant products are summed up,

wherein, when an acoustic effect is applied to an original tone by convoluting the original tone with impulse response data of an acoustic effect tone, the method further comprises;

implementing the FIR filter algorithm to convolute the original tone with first half data of the impulse response data in the time domain of a unit of a sampling period; and

implementing a Fast Fourier Transform algorithm on second half data of the impulse response data in the frequency domain of a unit of a block, thereby convoluting the original tone with the second half data.

15. The method according to claim 14, wherein the first half data and the second half data of the impulse response data are set arbitrarily.