JP5941009B2 - Noise analysis apparatus and method - Google Patents

Description

  The present invention relates to a noise analysis apparatus and method, and more particularly, to quality evaluation of audio performance of a computer.

  In the manufacturing process of computer systems, which are a collection of a wide variety of precision parts, in order to satisfy customer satisfaction at a high level, quality assessment of individual parts and whether there are problems when incorporated into the system, etc. Is verified. One standard for evaluating the quality of an audio chip incorporated in a computer system is whether the noise level called pop noise or click noise is within an allowable range. In the present specification, “noise” simply refers to pop noise and click noise.

  Pop noise and click noise are considered to be noise caused by bias fluctuation at the time of on / off of an audio codec and a preamplifier as a sound generation source, at the time of switching to these power saving functions. Patent Document 1 discloses a technique for reducing pop noise. With reference to paragraph 0029, a generation mechanism of pop noise in a notebook computer is described. Patent Document 2 describes an invention relating to an electronic musical instrument that analyzes an original waveform such as speech and feeds back an analysis result to a waveform synthesizer.

JP 2010-016795 A JP 2004-013179 A

  Conventionally, the quality evaluation of the audio performance of a computer is a qualitative evaluation in which an operator listens and judges with an ear. Therefore, a pop sound that seems to be within an allowable range for a certain operator may be heard as noise by a different operator. In qualitative evaluation, there are individual differences, which may allow the shipment of a computer system that incorporates an audio chip to be corrected, and conversely, an expensive chip may be incorporated due to too strict evaluation. .

  If quantitative evaluation becomes possible, it is possible to eliminate operator intervention in audio performance quality evaluation, so that a computer system with a certain quality can be efficiently produced, and customer satisfaction may be improved. .

  It is an object of the present invention to enable quantitative evaluation in quality evaluation related to computer pop noise.

In order to achieve the above object, an apparatus according to the present invention provides an approximate value for comparison that represents an approximate value of a minimum audible range indicating a minimum level curve in an equal loudness curve by a logarithmic ratio of an effective value of a maximum output level and a noise level. a threshold setting means for setting a threshold value and unit conversion means for performing unit conversion, a margin value to be set on the basis of the waveform data obtained by generating a pseudo noise, in addition to the minimum audible range after unit conversion to, Input means for inputting noise generated by an external evaluation sample, and a numerical value for comparison that represents the noise level for each frequency as a logarithmic ratio between the effective value of the maximum output level and the noise level based on the waveform analysis data of the input noise comprises a digitizing means for digitizing, based on said numerical for the comparison threshold, and evaluation means for evaluating the inputted noise, to the margin It is characterized by being set by a sensory test to determine the allowable whether noise said artificially generates heard operator.

Further, the method according to the present invention converts the approximate value of the minimum audible range indicating the lowest level curve in the equal loudness curve into a comparative value approximated by a logarithmic ratio between the effective value of the maximum output level and the noise level. A unit conversion step, a threshold value setting step for setting a threshold value in addition to the minimum audible range after unit conversion, a margin value set based on waveform data obtained by generating pseudo noise, and an external evaluation sample A numerical value that quantifies the noise level for each frequency into a numerical value for comparison based on the logarithmic ratio between the effective value of the maximum output level and the noise level based on the input process of inputting the generated noise and the waveform analysis data of the input noise. of a step, on the basis of said numerical for the comparison threshold, and an evaluation step for evaluating the input noise, the margin value, the pseudo The produced was noise, characterized in that it is set by a sensory test to determine whether the operator is listening allowable.

  According to the present invention, quantitative evaluation can be performed in quality evaluation regarding pop noise of a computer.

It is a figure which shows the hardware structural example (the 1) of embodiment of this invention. It is a figure which shows the hardware structural example (the 2) of embodiment of this invention. It is a block diagram which shows the function structure of embodiment of this invention. It is a figure which shows an equal loudness curve. In the embodiment of the present invention, it is a figure showing the result of qualitatively evaluating the degree of influence on the user when the noise level [dB reference] of the test tone is changed. 5 is a graph showing an example of threshold waveform data in the embodiment of the present invention.

  Hereinafter, embodiments according to the present invention will be described with reference to the drawings.

  The present embodiment is divided into a phase in which a threshold value for evaluating noise is set and a phase in which noise generated by the evaluation sample 11 is evaluated using the set threshold value. FIG. 1 shows an example of a hardware configuration that can be used in a phase for setting a threshold. FIG. 2 shows an example of a hardware configuration that can be used in the noise evaluation phase.

  In FIG. 1, a test tone generating computer 10 may be a personal computer that can take various forms such as a desktop type and a notebook type. The computer 10 for generating a test tone includes a terminal that can use a headphone of a type that terminates with an output of 5 mW and a resistance value of 32Ω, but this configuration is an example, and the present invention is not limited to this example. .

  On the other hand, the headphones 20 have a maximum output level of 100 [dBS PL / mW] as an example. In the specific example illustrated here, the maximum output output from the test tone generating computer 10 to the headphones 20 can be approximated to about 107 [dB SPL]. Note that this approximate value may be changed when headphones under other conditions are used.

  In FIG. 2, an evaluation sample 11 is a personal computer that is an object to be evaluated regarding pop noise and click noise. An audio chip may be used. The output specification of the headphone or line output of the evaluation sample 11 and the headphone output specification of the test tone generating computer 10 need to be the same. Therefore, in this example, the terminal of the evaluation sample 11 is a terminal of 5 mW output at the end of 32Ω. The evaluation sample 11 generates sample noise by artificially causing a bias fluctuation by turning on / off the audio codec, shifting to or returning from the power saving function.

  The input cable 21 connects the headphone or line output terminal of the evaluation sample 11 and the analog input of the waveform analyzer 12. A termination resistor is connected to the analog input section. For both headphones and line output, the termination resistance value can be changed according to the output specification of each evaluation sample. This example (in the case of headphones) terminates at 32Ω.

  The waveform analyzer 12 has a function of converting an input transient audio transient waveform in the time domain into a frequency domain and digitally outputting it. Any analysis device that can realize such a function is not limited. Specifically, the input audio may be converted into the frequency domain by performing the following processing.

  First, the DC component is removed from the input noise signal, and only the AC component is extracted (that is, only the bias fluctuation is extracted). Next, set the input range to measure the noise at a certain appropriate input range and use a bandpass filter based on it (the frequency range of the human ear is not flat, so the low frequency range of the noise, To attenuate high-frequency components). Next, the output signal of the band pass filter is analog / digital converted, and the converted signal is subjected to waveform analysis and converted to the frequency domain. Data converted to the frequency domain is output as waveform analysis data.

The waveform analyzer 12 outputs the waveform analysis data as a numerical value of the noise level for each frequency logarithmically divided within a range of 20 Hz to 20 kHz, for example. The numerical value of the noise level is represented by a logarithmic ratio (dB reference) between the effective value of the maximum output level of the evaluation sample 11 and the noise level. dB reference is obtained from the following equation.
dB reference = 20 * log ₁₀ (V _N / V _MAX )
V _N : Noise level
V _MAX : Maximum output level of the evaluation sample

  An output cable 22 that connects the waveform analyzer 12 and the waveform analysis computer 13 is a digital interface such as a USB.

  As the waveform analysis computer 13, for example, a personal computer that can take various forms such as a desktop type and a notebook type can be used. An interface (digital input terminal such as USB) to which the output cable 22 can be connected is provided. In FIG. 2, the waveform analysis device 12 and the waveform analysis computer 13 are separated, but the present invention is not limited to this configuration example. The waveform analysis computer 13 can also be configured to have the functions of the waveform analysis device 12.

  FIG. 3 shows a functional configuration of the present embodiment. The test tone generation computer 10, the waveform analysis device 12, and the waveform analysis computer 13 function as the noise analysis means 1 of FIG. 3 by information processing by a software program using hardware as shown in FIGS. To do. The noise analysis unit 1 includes a unit conversion unit 101, a threshold setting unit 102, an input unit 103, a numerical unit 104, and an evaluation unit 105.

  The unit conversion means 101 and the threshold setting means 102 can be implemented using, for example, the test tone generating computer 10. The unit conversion means 101 sets a threshold setting criterion based on the minimum audible range indicating the lowest level curve in the equal loudness curve.

  The equal loudness curve is well-known information standardized by international standardization standards. For example, in the ISO 226: 2003 equal loudness curve as shown in FIG. 4, the minimum level curve (minimum audible range curve) is approximated as follows.

  The equal loudness curve is obtained by measuring the sound pressure level at which the loudness and noise level of human hearing are the same for each different frequency and connecting them with contour lines. The unit is dB SPL (Sound Pressure Level) as described above. However, the unit of the noise level that is digitized by the digitizing means 104 described later is dB reference. Therefore, since simple comparison is not possible, the unit conversion means 101 unifies units by performing predetermined calculations in the minimum audible range.

Unit unification here is conversion from dB SPL to dB reference. In the specific example given as an example of the headphones 20 in the description of FIG. 1, the maximum output output from the test tone generating computer 10 to the headphones 20 can be approximated to about 107 [dB SPL]. Since this maximum output 107 [dB SPL] is a value when the output of the evaluation sample is maximum, in the case of using the headphones 20 of this specific condition,
107dB SPL ≒ 0dB reference
Can be considered. At this time, the unit system can be converted to dB reference by subtracting 107 [dB SPL] from the minimum audible range ATH (f). The minimum audible range ATH (f) ′ after unit conversion is calculated by the following equation.
ATH (f) '[dB reference] = ATH (f)-107 [dB SPL]

  ATH (f) ′ represents a minimum audible curve when the headphone 20 of the specific example is used. In practice, however, even higher noise levels are often acceptable. Therefore, the noise threshold is obtained by adding some margin value to ATH (f) ′. In FIG. 3, the threshold setting means 102 sets a predetermined margin value as a threshold in addition to the minimum audible range after unit conversion.

  The margin value is set by a sensory test (qualitative evaluation) in which an operator listens to the noise generated by the test tone generation computer 10 through the headphones 20 and judges the noise level (FIG. 1). The artificially generated noise is called a test tone. In other embodiments, the test tone may be generated by the evaluation sample 11.

  The test tone is obtained, for example, by combining several sine waves to create waveform data close to the waveform data of pop noise. The test tone thus obtained is qualitatively evaluated by changing the noise volume in a frequency band of 20 Hz to 20 kHz. FIG. 5 shows the result of qualitative evaluation of the degree of influence on the user when the test tone is generated from the test tone generating computer 10 and the noise level [dB reference] of the test tone is changed.

Based on the result of this qualitative evaluation, for example, when it is determined that the margin value is 30 [dB reference], the noise threshold is calculated by the following equation.
Noise threshold = ATH (f) '+ 30 [dB reference]
Further, the noise threshold is a threshold line in FIG. 6 when viewed from the waveform data graph. FIG. 6 is a graph showing an example of threshold waveform data. In FIG. 6, the minimum audible curve of the equal loudness curve is converted from dB SPL to dB reference, and a margin value of 30 dB is given.

  Based on the threshold value set by the above criteria, the noise of the evaluation sample 11 is analyzed. The input unit 103 inputs the noise of the evaluation sample 11 to the noise analyzer 1. The digitizing means 104 digitizes the inputted noise into a numerical value for comparison (FIG. 3).

As described once with reference to FIG. 2, the digitizing means 104 outputs the waveform analysis data as a numerical value of the noise level for each frequency logarithmically divided in the range of 20 Hz to 20 kHz, for example. The numerical value of the noise level is represented by a logarithmic ratio (dB reference) between the effective value of the maximum output level of the evaluation sample 11 and the noise level. dB reference is obtained from the following equation.
dB reference = 20 * log ₁₀ (V _N / V _MAX )
V _N : Noise level
V _MAX : Maximum output level of the evaluation sample

  The evaluation unit 105 can be realized by, for example, the waveform analysis computer 13 that stores the threshold set by the threshold setting unit 102. The evaluation unit 105 performs quantitative evaluation of noise input from the evaluation sample 11 based on the threshold value and the numerical value for comparison.

  According to the embodiment, quantitative evaluation can be performed in quality evaluation related to pop noise of a computer. In addition, since quantitative evaluation is possible, operator intervention can be eliminated in quality evaluation of audio performance. In addition, a computer system with a certain quality can be efficiently produced, and customer satisfaction can be improved. Further, when the quantitative evaluation method as described above is incorporated as one step in the manufacturing process of the computer system, a problematic device can be easily extracted. The efficiency of the manufacturing process becomes possible, and it becomes easy to manufacture a high-quality product.

DESCRIPTION OF SYMBOLS 1 Noise analysis means 10 Test tone generation computer 11 Evaluation sample 12 Waveform analysis apparatus 13 Waveform analysis computer 20 Headphone 21 Input cable 22 Output cable 101 Unit conversion means 102 Threshold setting means 103 Input means 104 Numerical means 105 Evaluation means

Claims (3)

Unit conversion means for converting the approximate value of the minimum audible range indicating the lowest level curve in the equal loudness curve into the approximate value for comparison represented by the logarithmic ratio of the effective value of the maximum output level and the noise level ;
A threshold value setting means for setting a threshold value in addition to a minimum audible range after unit conversion, a margin value set based on waveform data obtained by generating pseudo noise;
Input means for inputting noise generated by an external evaluation sample;
A digitizing means for quantifying the noise level for each frequency into a numerical value for comparison expressed by a logarithmic ratio of the effective value of the maximum output level and the noise level based on the input noise waveform analysis data;
Evaluation means for evaluating the input noise based on the comparison numerical value and the threshold;
Equipped with a,
The noise analysis apparatus according to claim 1, wherein the margin value is set by a sensory test in which an operator listens to the pseudo-generated noise and determines whether it is acceptable .   2. The noise analysis according to claim 1, wherein the numerical value unit represents the comparison numerical value for each frequency by a logarithmic ratio between an effective value of the maximum output level of the evaluation sample and a noise level. apparatus. A unit conversion step for converting the approximate value of the minimum audible range indicating the lowest level curve in the equal loudness curve to the approximate value for comparison represented by the logarithmic ratio of the effective value of the maximum output level and the noise level ; and
A threshold value setting step for setting a threshold value in addition to a minimum audible range after unit conversion, a margin value to be set based on waveform data obtained by generating pseudo noise;
An input process for inputting noise generated by an external evaluation sample;
A quantification process that quantifies the noise level for each frequency based on the input noise waveform analysis data into a numerical value for comparison that is expressed as a logarithmic ratio between the effective value of the maximum output level and the noise level
An evaluation step for evaluating the input noise based on the comparison numerical value and the threshold value;
Equipped with a,
The noise analysis method according to claim 1, wherein the margin value is set by a sensory test in which an operator listens to the pseudo-generated noise and determines whether it is acceptable .

Images