Digital signal processing device and method for controlling digital signal processing device
The digital signal processing device addresses the limitation of fixed delay times by using buffer memories and SDRAMs for burst transfer, enabling varied delay times that enhance audio quality through combined burst and sampling cycle-based delays.
Patent Information
- Application Number
- JP2020172560
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2020-10-13
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2040-10-13
AI Technical Summary
Existing digital signal processing devices limit delay time of sound waveform data to multiples of burst length when using DRAM burst transfer, leading to periodicity and strong correlation, which hinders high-quality audio effects.
A digital signal processing device utilizing a first and second buffer memory, SDRAMs for burst transfer, and a delay unit to generate audio data with delay times that combine multiples of burst length and sampling cycles, allowing for varied delay times.
Enables the generation of audio data with diverse delay times, including multiples of burst length and sampling cycles, thereby achieving high-quality audio effects like reverb by synthesizing original and delayed signals.
Smart Images

Figure 0007591901000001 
Figure 0007591901000002 
Figure 0007591901000003
Abstract
Description
[Technical field]
[0001] The present invention relates to a digital signal processing device and a method for controlling a digital signal processing device. [Background technology]
[0002] Patent Document 1 describes an address circuit for performing a virtual shift in a digital signal processing device. The address circuit combines an index register with an auxiliary index register, sets the same value as the lower value of the index register in the auxiliary index register, retrieves the upper address from the index register and the lower address from the auxiliary index register to perform normal index modification, and sets a value obtained by adding the lower value of the index register and a pointer to the auxiliary index register to perform address modification for the virtual shift.
[0003] Patent Document 2 describes an address generating circuit having a first counter and a second counter. A clearing means selectively clears the contents of the second counter to zero by a count control signal for the first counter. An adder adds the output of the first counter and the output of the second counter.
[0004] Patent Document 3 describes a digital signal processing device for sound waveform data having a waveform signal processing unit and a memory access unit. The waveform signal processing unit performs waveform signal processing for adding effects to sound waveform data by M (M is an integer equal to or greater than M) time-division waveform signal processing. The memory access unit enables an external DRAM having K banks (K is an integer equal to or greater than M) corresponding to K time-division waveform signal processing for delaying sound waveform data in the process of waveform signal processing to start accessing the K banks at K mutually shifted timings, and when a write or read request is received from one of the K time-division waveform signal processing, the memory access unit writes or reads sound waveform data by outputting a control signal for writing or reading to the corresponding bank to the DRAM at a timing at which access to the bank can start. [Prior art documents] [Patent documents]
[0005] [Patent Document 1] Japanese Patent Application Publication No. 58-2935 [Patent Document 2] Japanese Patent Application Publication No. 62-57067 [Patent Document 3] JP 2003-108122 A Summary of the Invention [Problem to be solved by the invention]
[0006] In Patent Document 3, sound waveform data is delayed in order to add effects to the sound waveform data. However, when using DRAM burst transfer, the delay time of the sound waveform data is limited to a multiple of the burst length. If multiple delayed signals are all multiples of the burst length, a strong correlation appears due to their periodicity, making it impossible to achieve high quality effects.
[0007] An object of the present invention is to make it possible to generate audio data having a delay time that is a combination of a delay time that is a multiple of the burst length relative to a sampling period and a delay time in units of a sampling period. [Means for solving the problem]
[0008] The digital signal processing device of the present invention includes a first buffer memory for storing audio data, a second buffer memory for storing audio data, a DRAM for burst-transferring audio data of a burst length, and a , when converting the original sound into audio data a delay means for delaying audio data in sampling cycle units, and a control means for sequentially writing audio data one word at a time into said first buffer memory in said sampling cycle, controlling the audio data of said burst length to be burst-transferred from said first buffer memory to said DRAM, controlling the audio data of said burst length to be burst-transferred from said DRAM to said second buffer memory, and outputting audio data one word at a time from said second buffer memory to said delay means in said sampling cycle, wherein the delay time of the audio data output by said delay means is a plurality of times depending on the burst length of said DRAM. The above A delay time in sampling period units and the delay means The above It is determined in combination with a delay time in sampling cycle units.
[0009] The control method of the digital signal processor of the present invention includes a first buffer memory for storing audio data, a second buffer memory for storing audio data, a DRAM for burst-transferring audio data of a burst length, and a control circuit for controlling the audio data. , when converting the original sound into audio dataa delay means for delaying audio data in sampling period units, the method comprising the steps of: sequentially writing audio data one word at a time into said first buffer memory in said sampling period; controlling a burst transfer of audio data of said burst length from said first buffer memory to said DRAM; controlling a burst transfer of audio data of said burst length from said DRAM to said second buffer memory; and sequentially outputting audio data one word at a time from said second buffer memory to said delay means in said sampling period, wherein a delay time of the audio data output by said delay means is a plurality of times which depend on the burst length of said DRAM. The above A delay time in sampling period units and the delay means The above It is determined in combination with a delay time in sampling cycle units. Effect of the Invention
[0010] According to the present invention, it is possible to generate audio data having a delay time that is a combination of a delay time that is a multiple of the burst length relative to the sampling period and a delay time in units of the sampling period. [Brief description of the drawings]
[0011] [Figure 1] FIG. 1 is a diagram showing an example of the configuration of a digital signal processing device according to this embodiment. [Diagram 2] FIG. 2 is a diagram showing a method of accessing the first buffer memory. [Diagram 3] FIG. 3 is a diagram showing a method of accessing the second buffer memory. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0012] FIG. 1 is a diagram showing an example of the configuration of a digital signal processing device 100 according to this embodiment. The digital signal processing device 100 is, for example, an effect device that realizes reverb. When an original sound is generated, a direct sound of the original sound and a delayed sound are synthesized and reach the human ear. The direct sound is the sound of the original sound that reaches the human ear directly. The delayed sound is a plurality of delayed sounds with different delay times that are generated by the original sound being reflected by various objects. The digital signal processing device 100 realizes reverb by generating a plurality of delayed signals with different delay times for the audio data of the original sound and synthesizing the audio data of the original sound with the plurality of delayed signals.
[0013] The digital signal processing device 100 has a CPU 101, a program ROM 102, a work RAM 103, a bus 104, an analog / digital converter 105, a DSP 106, first and second buffer memories 107, a digital / analog converter 108, a bus 109, an SDRAM 110, an SDRAM 111, and a delay unit 114. The delay unit 114 has a shift register 112 and a selector 113. A method of controlling the digital signal processing device 100 will be described below.
[0014] Each of the first and second buffer memories 107 is capable of inputting and outputting 32-bit (1-word) audio data to and from the bus 109. The first and second buffer memories 107 are, for example, SRAMs (Static Random Access Memories) and can store audio data.
[0015] The SDRAMs 110 and 111 are synchronous dynamic random access memories, which are a type of DRAM (dynamic random access memory). The SDRAMs 110 and 111 are, for example, DDR3 SDRAMs.
[0016] The SDRAM 110 is capable of inputting and outputting 16-bit (1 word) audio data to and from the bus 109. The SDRAM 110 is capable of inputting and outputting 16-bit (1 word) audio data to and from the bus 109.
[0017] Each of the first and second buffer memories 107 can input and output 32 bits (1 word) of audio data to and from the SDRAMs 110 and 111 via the bus 109. The 16-bit audio data input and output by the SDRAM 110 is the upper 16 bits of the 32-bit audio data input and output by the first or second buffer memory 107. The 16-bit audio data input and output by the SDRAM 111 is the lower 16 bits of the 32-bit audio data input and output by the first or second buffer memory 107.
[0018] The SDRAMs 110 and 111 are capable of burst-transferring audio data with a burst length of two words or more to each of the first and second buffer memories 107 via the bus 109. That is, the SDRAMs 110 and 111 continuously transfer audio data of the number of words of the burst length at high speed to each of the first and second buffer memories 107 via the bus 109 based on one address information. The burst length is, for example, four or eight words. An example in which the burst length is eight words will be described below.
[0019] CPU 101 is a central processing unit. Program ROM (Read Only Memory) 102 stores programs. Work RAM (Random Access Memory) 103 functions as a work area for CPU 101. CPU 101 loads the programs stored in program ROM 102 into work RAM 103, and controls DSP 106 by executing the programs loaded into work RAM 103. DSP 106 is a digital signal processor, and is a type of control unit.
[0020] An analog / digital converter 105 receives an analog audio signal of an original sound from a microphone or the like, and converts the analog audio signal into digital audio data of one word (32 bits) at a sampling period. The audio data is musical tone data or voice data.
[0021] The DSP 106 writes the digital audio data converted by the analog / digital converter 105 into the first buffer memory 107 one word at a time in a sampling period. Note that the DSP 106 may input digital audio data from a musical tone generating device, an audio receiving device, or the like, instead of the analog / digital converter 105.
[0022] Next, the DSP 106 controls burst transfer of audio data of the burst length word number (8 words) from the first buffer memory 107 to the SDRAMs 110 and 111. The DSP 106 writes the audio data of the burst length word number (8 words) to addresses indicated by the write pointers of the SDRAMs 110 and 111.
[0023] Next, the DSP 106 performs control so that audio data of the number of words of the burst length (8 words) is burst-transferred from the SDRAMs 110 and 111 to the second buffer memory 107. At this time, the DSP 106 performs control so that the audio data of the number of words of the burst length (8 words) is read from addresses indicated by the read pointers of the SDRAMs 110 and 111, and is burst-transferred. The DSP 106 writes the audio data of the number of words of the burst length (8 words) to the second buffer memory 107 by burst transfer.
[0024] The difference between the address indicated by the write pointer and the address indicated by the read pointer of the SDRAMs 110 and 111 corresponds to the delay time of the delayed signal relative to the original sound. This delay time is a multiple of the burst length (a multiple of 8) relative to the sampling period.
[0025] It is possible to generate delay signals using SRAMs instead of the SDRAMs 110 and 111, but a large-capacity SRAM is required to generate delay signals with long delay times. SRAMs have the disadvantage of being more expensive than the SDRAMs 110 and 111. Therefore, in this embodiment, the SDRAMs 110 and 111 are used to generate delay signals, thereby reducing costs. In addition, by using burst transfers of the SDRAMs 110 and 111, a large number of delay signals can be generated at high speed in real time.
[0026] Next, the DSP 106 reads out the audio data transferred to the second buffer memory 107 in units of one word at a sampling period, and outputs the read audio data to the delay section 114 .
[0027] The delay section 114 has a shift register 112 and a selector 113, and delays the audio data read from the second buffer memory 107 by a sampling period, and outputs the delayed audio data to the DSP 106 as a delayed signal.
[0028] The shift register 112 includes a plurality of registers for storing audio data read from the second buffer memory 107, and shifts the audio data stored in the plurality of registers in units of a sampling period. The selector 113 selects at least one of the output data of the plurality of registers, and outputs the selected output data to the DSP 106 as a delayed signal with respect to the original sound. The selection of the selector 113 corresponds to the delay time of the delayed signal with respect to the original sound.
[0029] Next, the DSP 106 synthesizes the audio data of the original sound converted by the analog / digital converter 105 with the above-mentioned delayed signal, and outputs the synthesized audio data to the digital / analog converter 108. Note that the synthesis does not necessarily have to be performed by the DSP 106, and may be performed by a device external to the DSP 106.
[0030] The digital / analog converter 108 converts one word (32 bits) of digital audio data into an analog audio signal at a sampling period, and outputs the analog audio signal to the audio system. The audio system has an amplifier and a speaker, amplifies the audio signal with the amplifier, and produces the amplified audio signal through the speaker. This produces a sound with a reverb effect added to the original sound.
[0031] 2 is a diagram showing a method of accessing the first buffer memory 107. The first buffer memory 107 is a ring buffer capable of storing 8 words of audio data. Sampling periods T0 to T8 are the sampling periods of the analog / digital converter 105, respectively.
[0032] In a sampling period T0, the DSP 106 writes one word of audio data W(0) converted by the analog / digital converter 105 to a first address in the first buffer memory 107.
[0033] Sampling period T1 is one sampling period after sampling period T0. In sampling period T1, the DSP 106 writes one word of audio data W(1) converted by the analog / digital converter 105 to a second address in the first buffer memory 107.
[0034] Sampling period T2 is a sampling period that is two sampling periods after sampling period T0. In sampling period T2, the DSP 106 writes one word of audio data W(2) converted by the analog / digital converter 105 to a third address in the first buffer memory 107.
[0035] Similarly, in sampling periods T3 to T6, the DSP 106 writes one word of audio data W(3) to W(6) converted by the analog / digital converter 105 to the fourth to seventh addresses of the first buffer memory 107, respectively.
[0036] Sampling period T7 is a sampling period that is seven sampling periods after sampling period T0. In sampling period T7, the DSP 106 writes one word of audio data W(7) converted by the analog / digital converter 105 to the eighth address of the first buffer memory 107.
[0037] The DSP 106 burst-transfers eight words of audio data W(0) to W(7) stored in the first buffer memory 107 to the SDRAMs 110 and 111 at a sampling period T7, and writes the eight words of audio data W(0) to W(7) to the SDRAMs 110 and 111.
[0038] Sampling period T8 (T0) is a sampling period that is eight sampling periods after sampling period T0. In sampling period T8 (T0), the DSP 106 writes (overwrites) one word of audio data W(8) converted by the analog / digital converter 105 to a first address of the first buffer memory 107.
[0039] Similarly, in sampling periods T9 to T15, the DSP 106 writes (overwrites) one word of audio data W(9) to W(15) converted by the analog / digital converter 105 to the second to eighth addresses of the first buffer memory 107, respectively.
[0040] Thereafter, at sampling period T15, the DSP 106 burst-transfers eight words of audio data W(8) to W(15) stored in the first buffer memory 107 to the SDRAMs 110 and 111, and writes the eight words of audio data W(8) to W(15) to the SDRAMs 110 and 111.
[0041] Thereafter, in a similar manner, the DSP 106 burst-transfers the audio data stored in the first buffer memory 107 to the SDRAMs 110 and 111 in units of 8 words every 8 sampling periods.
[0042] As described above, the DSP 106 controls so that, in sampling periods T0 to T8, the audio data converted by the analog / digital converter 105 is written sequentially into the first buffer memory 107 one word at a time, and so that, in eight sampling periods, audio data of a burst length is burst-transferred from the first buffer memory 107 to the SDRAMs 110 and 111.
[0043] 3 is a diagram showing a method of accessing the second buffer memory 107. The second buffer memory 107 is a ring buffer capable of storing 8 words of audio data. Sampling periods T0 to T3 are the sampling periods of the analog / digital converter 105, respectively.
[0044] The delay unit 114 has a shift register 112 and a selector 113. The shift register 112 includes seven registers 112a to 112g for storing audio data, and shifts the audio data stored in the seven registers 112a to 112g in sampling period units. The selector 113 selects either the audio data output by the second buffer memory 107 or the output data of the seven registers 112a to 112g.
[0045] In a sampling period T0, the DSP 106 burst-transfers audio data W(0) to W(7) of the burst length word number (8 words) stored in the SDRAMs 110 and 111 to the second buffer memory 107.
[0046] In Fig. 2, the DSP 106 writes audio data stored in the first buffer memory 107 to addresses indicated by the write pointers of the SDRAMs 110 and 111. In a sampling period T0 in Fig. 3, the DSP 106 reads audio data from addresses indicated by the read pointers of the SDRAMs 110 and 111. The difference between the addresses indicated by the write pointers and the read pointers of the SDRAMs 110 and 111 corresponds to the delay time of the delayed signal relative to the original sound. This delay time is a multiple of 8 of the sampling period (a multiple of the number of burst lengths).
[0047] In a sampling period T0, the DSP 106 reads one word of audio data W(0) stored in a first address of the second buffer memory 107, and outputs the read audio data W(0) to the shift register 112 and the selector 113. The audio data W(0) is stored in a register 112g of the shift register 112.
[0048] In a sampling period T1, the DSP 106 reads one word of audio data W(1) stored in the second address of the second buffer memory 107, and outputs the read audio data W(1) to the shift register 112 and the selector 113. The shift register 112 shifts the audio data stored in the registers 112a to 112g. The register 112f stores audio data W(0). The register 112g stores audio data W(1).
[0049] In a sampling period T2, the DSP 106 reads one word of audio data W(2) stored in a third address of the second buffer memory 107, and outputs the read audio data W(2) to the shift register 112 and the selector 113. The shift register 112 shifts the audio data stored in the registers 112a to 112g. The register 112e stores audio data W(0). The register 112f stores audio data W(1). The register 112g stores audio data W(2).
[0050] In a sampling period T3, the DSP 106 reads one word of audio data W(3) stored in a fourth address of the second buffer memory 107, and outputs the read audio data W(3) to the shift register 112 and the selector 113. The shift register 112 shifts the audio data stored in the registers 112a to 112g. The register 112d stores audio data W(0). The register 112e stores audio data W(1). The register 112f stores audio data W(2). The register 112g stores audio data W(3).
[0051] Similarly, in sampling periods T4 to T6, the DSP 106 reads out the audio data W(4) to W(6) stored in the second buffer memory 107, and outputs the read audio data W(4) to W(6) to the shift register 112 and the selector 113.
[0052] In a sampling period T7, the DSP 106 reads one word of audio data W(7) stored in the eighth address of the second buffer memory 107, and outputs the read audio data W(7) to the shift register 112 and the selector 113. The shift register 112 shifts the audio data stored in the registers 112a to 112g. The audio data W(0) to W(7) are stored in the registers 112a to 112g, respectively.
[0053] Thereafter, the digital signal processing device 100 repeats the above-mentioned processing for the sampling periods T0 to T7. The DSP 106 controls so that audio data of a burst length (8 words) is burst-transferred from the SDRAMs 110 and 111 to the second buffer memory 107 in 8 sampling periods, and outputs the audio data from the second buffer memory 107 to the delay unit 114 in sequence, one word at a time, in the sampling period.
[0054] The shift register 112 shifts the audio data stored in the seven registers 112a to 112g in units of a sampling period. Each of the seven registers 112a to 112b outputs the audio data stored therein to the selector 113.
[0055] The selector 113 selects either the audio data read from the second buffer memory 107 or the output data of the seven registers 112 a to 112 g in accordance with the delay time setting, and outputs a delayed signal relative to the original sound to the DSP 106 .
[0056] The selector 113 selects the audio data read from the second buffer memory 107 and outputs the selected audio data to the DSP 106, thereby generating a delay signal having a delay time that is a multiple of 8 of the sampling period.
[0057] In addition, the selector 113 selects the audio data output by the register 112g, and when it outputs the selected audio data to the DSP 106, it can generate a delay signal having a delay time of (a multiple of 8 of the sampling period) + (sampling period) × 1.
[0058] In addition, the selector 113 selects the audio data output by the register 112f, and when it outputs the selected audio data to the DSP 106, it can generate a delay signal having a delay time of (a multiple of 8 of the sampling period) + (sampling period) × 2.
[0059] In addition, the selector 113 selects the audio data output by the register 112e, and when it outputs the selected audio data to the DSP 106, it can generate a delay signal having a delay time of (a multiple of 8 of the sampling period) + (sampling period) × 3.
[0060] In addition, the selector 113 selects the audio data output by the register 112d, and when it outputs the selected audio data to the DSP 106, it can generate a delay signal having a delay time of (a multiple of 8 of the sampling period) + (sampling period) × 4.
[0061] In addition, the selector 113 selects the audio data output by the register 112c, and when it outputs the selected audio data to the DSP 106, it can generate a delay signal having a delay time of (a multiple of 8 of the sampling period) + (sampling period) × 5.
[0062] In addition, the selector 113 selects the audio data output by the register 112b, and when it outputs the selected audio data to the DSP 106, it can generate a delay signal having a delay time of (a multiple of 8 of the sampling period) + (sampling period) × 6.
[0063] In addition, the selector 113 selects the audio data output by the register 112a, and when it outputs the selected audio data to the DSP 106, it can generate a delay signal having a delay time of (a multiple of 8 of the sampling period) + (sampling period) × 7.
[0064] As described above, the delay unit 114 can generate not only a delay signal having a delay time that is a multiple of 8 of the sampling period (multiple of the burst length) but also a delay signal having a delay time in sampling period units. The delay time of the audio data output by the delay unit 114 is determined by a combination of a delay time in multiple sampling periods that depends on the burst length of the SDRAMs 110 and 111 and the delay time in sampling period units of the delay unit 114.
[0065] It should be noted that the delay unit 114 is not limited to using the shift register 112 and the selector 113. For example, the delay unit 114 may delay the audio data in sampling period units by using random access of an SRAM.
[0066] The DSP 106 inputs the delayed signal from the selector 113 and synthesizes the original audio data converted by the analog / digital converter 105 with the delayed signal. The delay unit 114 generates a plurality of delayed signals with different delay times by time-division processing, and the DSP 106 synthesizes the original audio data with the plurality of delayed signals, thereby achieving high-quality reverb.
[0067] If all the multiple delayed signals are multiples of the burst length of the sampling period, a strong correlation will appear due to the periodicity, and high-quality reverb will not be realized. According to this embodiment, the delay unit 114 can generate not only delayed signals having delay times that are multiples of the burst length of the sampling period, but also delayed signals having delay times in units of the sampling period, thereby realizing high-quality reverb.
[0068] It should be noted that the above-mentioned embodiments are merely examples of the implementation of the present invention, and the technical scope of the present invention should not be interpreted as being limited by these. In other words, the present invention can be implemented in various forms without departing from its technical concept or main features. [Explanation of symbols]
[0069] 100 Digital signal processing device 101 CPU 102 Program ROM 103 Work RAM 104 Bus 105 Analog / Digital Converter 106 DSP 107 Buffer Memory 108 Digital / Analog Converter 109 Bus 110,111 SDRAM 112 Shift Register 113 Selector 114 Delay section
Claims
1. a first buffer memory for storing audio data; a second buffer memory for storing audio data; a DRAM for burst-transferring audio data of a burst length; a delay means for delaying the audio data by a sampling period when an original sound is converted into audio data; control means for sequentially writing audio data one word at a time into said first buffer memory in said sampling period, controlling a burst transfer of the audio data of said burst length from said first buffer memory to said DRAM, controlling a burst transfer of the audio data of said burst length from said DRAM to said second buffer memory, and sequentially outputting the audio data one word at a time from said second buffer memory to said delay means in said sampling period; a delay time of the audio data output by said delay means being determined by a combination of a plurality of delay times in units of said sampling period which depend on said burst length of said DRAM and a delay time in units of said sampling period of said delay means.
2. The delay means is a shift register including a plurality of registers for storing the audio data, the shift register shifting the audio data stored in the plurality of registers in units of the sampling period; 2. The digital signal processing device according to claim 1, further comprising a selector for selecting any one of the output data of at least the plurality of registers.
3. 3. The digital signal processing apparatus according to claim 2, wherein said selector selects either the audio data output by said second buffer memory or the output data of said plurality of registers.
4. 4. The digital signal processing device according to claim 1, wherein the DRAM is an SDRAM.
5. a first buffer memory for storing audio data; a second buffer memory for storing audio data; a DRAM for burst-transferring audio data of a burst length; A method for controlling a digital signal processing device having a delay means for delaying audio data by a sampling period when an original sound is converted into audio data, comprising the steps of: writing audio data into said first buffer memory one word at a time in said sampling period; a step of controlling a burst transfer of audio data having the burst length from the first buffer memory to the DRAM; a step of controlling a burst transfer of audio data having the burst length from the DRAM to the second buffer memory; outputting audio data from said second buffer memory to said delay means in sequence, one word at a time, during said sampling period; a delay time of the audio data output by said delay means being determined by a combination of a plurality of delay times in units of said sampling period which depend on said burst length of said DRAM and a delay time in units of said sampling period of said delay means.
Citation Information
Patent Citations
Address circuit
JP1983002935A
Address generating circuit
JP1987057067A
Electronic musical instrument
JP1996050480A
Audio signal generating device
JP2000197194A
Digital signal processor for sound waveform data
JP2003108122A
