Novel hearing test system

Abstract

The invention discloses a software unit intended for hearing assessment, said software unit being arranged to perform a method comprising the steps of:-performing a reference frequency hearing test to establish a reference frequency for each ear of a test subject; and-performing an equal loudness hearing test for each ear wherein the step of performing the reference frequency hearing test involves: providing sound to the test subject at a set sound level starting from a first frequency positioned within a given frequency range to obtain or not obtain a hearing response, the set sound level selects the lowest test sound level in a range of-10 Phon to 50 Phon; changing the sound level at a first frequency; selecting a first frequency as a reference frequency if the test subject provides a hearing response at the selected lowest test sound level in the range of-10 Phon to 50 Phon, otherwise selecting another frequency in a given frequency range; and repeating the iterative process at a new frequency within a given frequency range until the software unit detects a frequency X for which the test subject provides a hearing response at the selected lowest test sound level within the range of-10 Phon to 50 Phon, or the frequency is a frequency having a lowest sound level hearing response from the test subject for a set frequency to be tested, and then selecting the frequency X as a reference frequency.

Description

Background Technology

[0001] The commonly used ISO standard hearing threshold test involves detecting a person's hearing threshold at only a few frequencies (typically around 6 to 10). The test results are recorded in an audiogram. While the test can be used to detect hearing loss by detecting elevated hearing thresholds, it does not reveal any information about hearing loss above those thresholds. Any hearing rehabilitation applied based on the test must rely on the assumption that hearing loss is linear at higher sound pressure levels and is more or less equal to the elevated threshold. The novel equal-loudness hearing test clearly reveals that this is not the case. The hearing ability of people experiencing hearing loss changes significantly at sound pressure levels above their hearing threshold.

[0002] Figure 1 This diagram shows equal-loudness hearing test data for individuals with significant high-frequency hearing loss at levels of 10 phon, 20 phon, 30 phon, and 40 phon. The bottom trace represents the lowest 10 phon level, which is slightly above or at the subject's hearing threshold at 800 Hz. The right side of the trace shows a significant increase in the hearing threshold starting slightly above 1 kHz. On the left side of the diagram, at lower frequencies, the traces are spaced apart by the expected 10 phon. At higher frequencies, starting at approximately 2 kHz and above, the spacing is significantly smaller than 10 phon. At 300 Hz, the 10 phon to 40 phon traces are almost 40 dB apart, corresponding to an increase in sound level. A slight increase in the 10 phon trace between 100 Hz and 1 kHz makes the spacing between the 20 phon trace slightly smaller than the other traces. This is caused by the slight increase in the hearing threshold affecting the lowest 10 phon level. At 20 phon and above, the increase disappears, and the spacing is exactly 10 phon, except for some small local deviations. At higher frequencies above 2 kHz, the hearing threshold increases significantly, with the 10-phon to 40-phon traces separated by only a few phons. This reveals that at, for example, 3 kHz, a mere increase of a few phons in sound pressure level would be perceived as a 40-phon increase by the subject. Clearly, once the sound pressure level exceeds the hearing threshold, the brain begins to fundamentally compensate for the increased threshold, and the ear / brain no longer functions like a linear device. This obviously overturns the traditionally accepted assumption that hearing loss is linear.

[0003] This hearing loss behavior presents significant problems if the elevated hearing threshold is compensated using current standard methods that rely on assumptions about linear hearing behavior. Using standard compensation practices, amplifying the measured elevated hearing threshold by approximately half to two-thirds results in insufficient amplification at lower sound levels and frequencies above approximately 2 kHz, while at higher sound levels, the sound is amplified too much. This is further demonstrated by the fact that their hearing, measured using ELHT and as... Figure 1The subjects shown had received professionally fitted hearing aids, but did not use them because they offered no benefit.

[0004] Under normal circumstances, the dynamic range compression at frequencies where the hearing threshold rises is less than [value missing]. Figure 1 The situation presented, however, showed that all subjects measured exhibited similar dynamic compression at frequencies where the hearing threshold was elevated. Referring to the collected measurement data, it is clear that when sounds are above the hearing threshold and therefore audible, the brain automatically compensates for the elevated hearing threshold at higher sound pressure levels. This compensation by the brain can be quite significant (e.g., Figure 1 (as shown) or smaller, but the compensation is always present to a significant degree.

[0005] In order to not only assess the presence of hearing loss but also acquire data to achieve proper hearing recovery, a device containing software that facilitates new testing methods is needed. Therefore, the novel equal-loudness hearing test (ELHT) applies several additional testing steps compared to the standard ISO hearing threshold test. The following description also applies to the devices, software, and methods used to perform the ELHT.

[0006] Beyond its most crucial aspect—achieving proper hearing restoration—ELHT utilizes the available dynamic range efficiently. Current practices (i.e., ISO standard hearing tests and compensation based on static amplification) consume a significant portion of the available dynamic range without any benefit. The static amplification method effectively exposes users to the risk of additional hearing loss due to adverse and unnecessary exposure to sound pressure levels higher than desired. If standard tests reveal a hearing threshold increase of approximately 70 dB at 3 kHz (as per [reference to...]),... Figure 1 In the case of individuals with hearing loss, the normal approach is to amplify the sound at that frequency by approximately half to two-thirds of the measured rise. In this case, the amplification would be approximately 35 to 45 dB, assuming 40 dB is used.

[0007] ELHT test results indicate that 40dB amplification is insufficient below 30 phon. For sounds below 30 phon, this is perfectly adequate amplification, but at higher levels, it is excessive. At approximately 70 phon, from... Figure 1 The ELHT data clearly shows that amplification is no longer needed, making the static 40dB amplification excessive. Therefore, why does hearing loss... Figure 1 It is not surprising that the people shown cannot get help through traditional methods.

[0008] Using the ELHT method and dynamic amplification that varies according to sound level, users do not experience any additional amplification at sound levels above 70 phon, which eliminates the risk of continued hearing loss caused by the unintentional and unnecessary generation of high sound levels due to hearing loss compensation.

[0009] Regarding the utilization of available dynamic range; as an example, personal sound applications in mobile devices with Bluetooth-connected earbuds or headphones already have a very limited dynamic range, constrained by both the headphone hardware and the Bluetooth connection. The digital lower limit will never be better than 16-bit resolution (i.e., 96dB dynamic range), and in practice, this digital lower limit is even smaller, compounded by hardware limitations. If static amplification already uses 40dB of that dynamic range, then at most only 56dB remains. If we assume the maximum output of the Bluetooth headphone hardware is 100dB, then no sound below 44dB can be reproduced; this is the hard digital resolution lower limit, not the noise lower limit. Literally, it's impossible to reproduce sounds softer than 44dB, meaning a considerable amount of softer sound will be completely inaudible, defeating the purpose of the system—to restore hearing for better intelligibility. The same applies to hearing aids, which typically operate with an even more limited bit depth. Summary of the Invention

[0010] The foundation of ELHT is perceived equal loudness. The ISO 226 standard provides equal loudness information at different sound levels and frequencies, which ELHT uses to assess hearing loss. The output from the ELHT to the real-time signal processing unit is gain table data. The real-time signal processing unit should include a corrected peak filter at each frequency where amplification is needed to restore hearing. The corrected peak filter should dynamically adjust its boost at the filter frequency based on the instantaneous sound pressure level at that frequency. The gain table data output from the ELHT contains corrected gain information for all actual sound pressure levels at each frequency.

[0011] The key point of the method according to the invention is to establish at least one reference frequency for each ear. In its most general aspect, the invention relates to a software unit intended for use in hearing assessment, the software unit being configured to perform a method comprising the steps of:

[0012] - Perform reference frequency hearing tests to establish reference frequencies for each ear of the test subject; and

[0013] - Perform equal-loudness hearing tests on each ear.

[0014] The steps for performing a reference frequency hearing test involve: starting from a first frequency located within a given frequency range, providing sound to the test subject at a set sound level to obtain or not obtain a hearing response, wherein the set sound level is selected as the lowest test sound level within the range of -10 phon to 50 phon; changing the sound level at the first frequency; if the test subject provides a hearing response at the selected lowest test sound level within the range of -10 phon to 50 phon, then selecting the first frequency as the reference frequency; otherwise, selecting another frequency within the given frequency range; and repeating the iterative process with new frequencies within the given frequency range until the software unit detects a frequency X for which the test subject provides a hearing response at the selected lowest test sound level within the range of -10 phon to 50 phon, or which frequency is the frequency that has the lowest sound level hearing response from the test subject for the set frequency to be tested, and then selecting that frequency X as the reference frequency.

[0015] It should be noted that the present invention relates to a software unit intended for use in hearing assessment, but may also provide correction for one or more types of hearing impairment.

[0016] The first ELHT step aims to establish reference frequencies for equal loudness evaluation. One reference frequency is needed for each ear. The frequency may be the same for both the left and right ears, but in some cases, it is beneficial to use two different frequencies. The perceived loudness at the reference frequencies at different sound levels is used to balance the perceived loudness at all other frequencies in the test.

[0017] The second ELHT step balances the perceived loudness at a reference frequency between the left and right ears. Since all measurements are balanced against the sensed sound level at the reference frequency, it is crucial that the perceived levels in the left and right ears are equal at all sound pressure levels. If they differ, the overall spatial perception of sound will be skewed, leading to a decline in spatial hearing ability.

[0018] This second step is not mandatory, but it is preferred to be performed according to the invention.

[0019] The third step of ELHT is the equal loudness test. The perceived loudness at a reference frequency (one frequency per ear) is compared with multiple test frequencies appropriately distributed across a frequency range at multiple appropriate sound pressure levels above the hearing threshold.

[0020] Regarding steps four, five, and six described below, it should be noted that these steps are optional according to the invention. Similarly, the focus of the method according to the invention is to establish at least one reference frequency for each ear.

[0021] The fourth ELHT step transposes the equal loudness data measured in phons to sound pressure level (SPL). The output from the third ELHT step is in phons. SPL data is essential for efficient signal processing in hearing recovery devices such as earplugs and hearing aids. SPL is a physical unit typically measured in dBSPL, which is the logarithmic unit of the reference level using 20 micropascals of sound pressure, while phons are logarithmic units of perception. The perceived level measured in phons is based on human perception of sound loudness.

[0022] The fifth ELHT data post-processing step begins by adding missing hearing correction data. The measurements performed in step 3 only cover a limited dynamic range, typically 10 phon to 60 phon. Correct hearing correction requires data for the entire relevant dynamic range of 0 dBSPL to 110 dBSPL. The correction gain data for the missing portion of the dynamic range is calculated by polynomial extrapolation of the available measurement range at all frequencies. The next step involves selecting appropriate correction frequencies where correction gain should be applied. Frequencies where no correction gain is needed based on measurements are discarded. Frequencies where hearing loss is too severe for correction to be feasible are also discarded. This typically occurs at the highest frequencies of 9600 Hz and 12800 Hz, where age-related hearing loss may make these frequencies inaudible, so applying excessive gain would be pointless or even harmful. Finally, the remaining correction gain data is analyzed to find peaks in the hearing loss data suitable for correction. Due to the bandwidth of the correction peak filter, a peak frequency typically covers a frequency range, which usually eliminates the need to boost adjacent frequencies closest to the center frequency. When an appropriate frequency is selected, the last two steps include limiting the maximum gain to avoid acoustic feedback in the hearing correction device and limiting the gain, thereby limiting the maximum sound output at the highest sound pressure level.

[0023] The sixth ELHT step adjusts the overall amplification. Amplification from adjacent frequencies interacts and adds together, resulting in additional gain. The correction filter bandwidth, and consequently the amplification bandwidth, must be wide to achieve a good time-domain response with limited energy dispersion (i.e., ringing). Therefore, a wide bandwidth leads to overlap between adjacent adjustment bands. To correct the overall gain to equal the measured desired gain, optimization must be performed. This optimization is called gain sail optimization because the applied gain resembles a sail in a three-dimensional graph, where frequency, sound level, and amplification are on the axes. See [link to relevant documentation]. Figure 6 . Detailed Implementation

[0024] The ELHT system uses a novel hearing test to assess hearing loss across a wide dynamic range, significantly different from current ISO standard hearing threshold tests that only provide information about hearing thresholds. ELHT is based on the perceived loudness of a reference sound compared to the test sound. The test sound level is adjusted until the test subject perceives its loudness as equal to that of the reference sound. Typically, the reference and test sounds are played back alternately at an alternating frequency of approximately 0.5 Hz. The alternation frequency can obviously be changed to higher or lower frequencies. A feasible range would be 0.25 Hz to 1 Hz. The alternation between the two contrasting sounds can be automatic or manually controlled. The loudness levels used are based on the levels at different frequencies and sound pressure levels according to the ISO 226 standard. The test and reference sounds can be pure sine waves, or other types of sounds, with bandwidth limited to the desired test band. Composite test signals with a wider spectrum than pure sine waves can provide more information about hearing ability within the test band and are particularly useful in tests with fewer test bands.

[0025] The hearing threshold tests typically performed use pure sine tones, which can be a serious problem for individuals with tinnitus. With tinnitus, it can be difficult to distinguish the perceived tinnitus tone from the test tone generated to assess hearing ability, leading to confusion and inaccurate test results.

[0026] To improve accuracy and user-friendliness, vibrato is used in ELHT, a favorable alternative to pure sine waves. While many other types of multi-frequency tones and noise can be used, vibrato offers several beneficial properties. First, it is impossible to confuse vibrato with tinnitus. Vibrato sounds significantly different from the pure sine waves experienced by someone with tinnitus. Scientifically evaluated, vibrato produces very similar hearing threshold test results compared to pure sine waves. The bandwidth of vibrato extends just above and below the test frequency, thus making it less sensitive to local troughs and peaks in the frequency response generated by the testing equipment and the user's ear and ear canal. This may be particularly relevant at higher frequencies, where peaks and troughs are more pronounced. Finally, especially for equal-loudness testing, determining the loudness of vibrato is easier than determining the loudness of pure sine waves, making equal-loudness comparisons easier. The appropriate range for vibrato modulation is 1% to 10%, with ELHT typically using 5%. The standard ELHT modulation frequency is 15 Hz, with a useful range of 5 Hz to 30 Hz.

[0027] Figure 1The equal loudness test results shown use 81 frequencies in the range of 100 Hz to 10 kHz. ELHT does not require so many test frequencies. The number of frequencies can be any number between 1 and the presented 81, or even more. Preferably, the test frequencies should be 100 Hz, 200 Hz, 400 Hz, 800 Hz, 1200 Hz, 1600 Hz, 2400 Hz, 3200 Hz, 4800 Hz, 6800 Hz, 9600 Hz, and 12800 Hz. These frequencies are evenly distributed across the frequency range relevant to hearing recovery. These frequencies are more closely spaced at higher frequencies, where optimal correction requires higher resolution.

[0028] ELHT is performed in 10-Phon sound pressure level intervals, where the lowest test level should be slightly above the subject's hearing threshold at the selected reference frequency. Any other sound pressure level granularity can be used, but closer intervals than 10-Phon offer little benefit in terms of accuracy and are significantly more strenuous for the subject. Wider intervals are occasionally used during ELHT testing to find particular hearing threshold limits more quickly. In such cases, a 20-Phon interval is used. Again, any granularity can be used, but the 10-Phon basis strikes a good trade-off between accuracy, user-friendliness, and speed.

[0029] Figure 2 The graphical user interface (GUI) from the ELHT software application is shown. The figure illustrates step 1, where the GUI provides support functionality for setting up and generating calibration tables for the DSP hearing correction device.

[0030] If a subject has low or no hearing loss at the selected reference frequency, the lowest ELHT level is typically set at 10 phon. If a subject experiences hearing loss at the selected reference frequency, the lowest level can be increased. The lowest ELHT level should be slightly higher than the subject's hearing threshold at the reference frequency, i.e., 0 to 10 phon higher. At least one higher sound pressure level is required to assess hearing ability. More sound pressure test levels can better assess a subject's hearing ability. Generally, four test levels with a 10 phon interval produce acceptable results. Testing at more levels significantly increases the burden on the subject but will provide a better understanding of hearing ability across a larger dynamic range.

[0031] Step 1, Selection of Reference Frequency

[0032] The purpose of the reference selection process is to find a reference frequency for each ear, where hearing ability is as normal as possible without significant hearing loss. Each ear is tested individually to establish an appropriate frequency, one frequency per ear. The reference frequency can be the same for both ears, or it can be different for each ear if necessary.

[0033] The goal is to position the reference frequency as close as possible to the middle of the frequency range being tested. It is easier to compare the perceived loudness of two sounds if their frequency content is closer. Age-related hearing loss typically occurs at higher frequencies, while hearing ability remains relatively intact in the lower frequency range. Hearing loss due to exposure to high sound levels also typically occurs at higher frequencies. Therefore, it is advantageous to begin the reference frequency selection test with an 800Hz or 1200Hz vibrato. 1200Hz is closer to the middle of the test range than 800Hz, but some hearing loss occasionally occurs at 1200Hz, making 800Hz more feasible in some cases. Acceptable reference frequencies are 400Hz, 800Hz, 1200Hz, 1600Hz, 2400Hz, 3200Hz, and 4800Hz. In some cases, 200Hz and 6800Hz can be used, but lower accuracy is expected due to the increased distance between the reference frequency and some of these test frequencies.

[0034] Higher frequencies within the acceptable range may be useful for individuals with what is commonly known as "cookie bite" hearing loss (i.e., loss of mid-frequency hearing but retention of hearing at the lowest and highest frequencies). Generally, for individuals with severe hearing loss due to age or exposure to high sound levels, lower frequencies are typically used when there is significant loss at 1200 Hz.

[0035] The following steps were performed individually for each ear, and the entire test was conducted using vibrato. The reference frequency tests began at 60 phon and 1200 Hz. Figure 3 The test panel GUI is shown. Whenever a sound is heard, the subject should press the large button on the GUI panel. The test frequency sequence is 1200Hz, 800Hz, 1600Hz, 2400Hz, 400Hz, and 3200Hz.

[0036] If the frequency sequence is not heard at 60 phon, the test will then proceed to 200 Hz and finally 6800 Hz at 60 phon. If the frequency sequence is still not heard, the level will be increased to 70 phon, and the sequence will be tested again. If the frequency sequence is still not heard, the hearing loss is too severe, and the test will be terminated.

[0037] When a frequency in the sequence is heard, stop at that frequency and test at 20 phons below the current level. If that frequency is heard, test at another 20 phons below. Continue until 10 phons is reached and heard, then stop at that frequency and use that frequency as a reference frequency. For test subjects with normal hearing at the test frequency, the level sequence will be 60 phons, 40 phons, 20 phons, 10 phons, and stop at that level if the test tone is heard at all levels. If the hearing threshold is between 20 phons and 30 phons, the sequence will be 60 phons, 40 phons, 20 phons, 30 phons, and then stop. 30 phons will be the lowest level at which the test frequency is heard.

[0038] If the lowest level of the test frequency is heard to be higher than 10 phon, then start testing the next frequency in the sequence from the lowest level of the current frequency.

[0039] When you hear one of the following frequencies in the sequence, test at that frequency at a level below 10 phons. Continue until you reach and hear 10 phons, then stop at that frequency and use it as the reference frequency, or stop earlier at the frequency heard at the lowest level and use that frequency. If you hear several frequencies at a particular level, use the first audible frequency in the sequence as the reference frequency.

[0040] The perceived loudness of other test frequencies is balanced using one or more selected reference frequencies at all sound levels; that is, one frequency per ear at all levels, which may or may not be the same for both ears.

[0041] Step 2: Balance the reference frequency levels for the left and right ears.

[0042] The reference frequency established in step 1 (one reference frequency for reaching the ear) will be used as the reference frequency to balance all other test frequency levels. The perceived loudness of the other test frequencies should match the loudness of the reference frequency at all sound levels.

[0043] Importantly, the perceived loudness of the reference frequency was identical in both the left and right ears across all test levels. If it weren't, the entire frequency range where the loudness is balanced against the reference loudness would deviate towards one ear. Furthermore, if the loudness balance deviates differently across levels, the spatial sound field would bounce back and forth according to the level.

[0044] Figure 4The test GUI is shown. Left-right balance is controlled by sliders and buttons on the panel. If the reference frequency is the same for both the left and right ears, inter-ear balance is tested at each sound level that will be used in the test. Gain should only be used to balance the perceived loudness levels of the left and right ears, not attenuation. The loudest frequency is the frequency at which hearing is optimal, and therefore should be used to balance loudness; that is, apply gain to the ear that hears the softest sound.

[0045] If the reference frequencies are different for the left and right ears, interaural balance must be tested at both frequencies for each sound level used in the test. As an example, use 1200Hz as the reference frequency for the left ear and 800Hz as the reference frequency for the right ear. First, balance the left ear loudness at 800Hz with the left ear reference loudness at 1200Hz, so that the perceived loudness in the left ear is the same at both frequencies. Similarly, balance 1200Hz with the reference frequency 800Hz in the right ear. Then, balance 1200Hz and 800Hz in the right ear with the left ear reference frequency of 1200Hz, and in the same way, balance the left ear with the right ear reference frequency of 800Hz. Apply gain to all frequencies in both ears so that the softer frequencies are balanced with the loudest frequencies; that is, no gain is applied to the loudest frequencies in either ear, and gain is applied to all other frequencies in either ear.

[0046] Step 3, equal loudness test

[0047] During the equal loudness test, the perceived loudness of each reference frequency is compared with the loudness of the test frequency at different sound levels. The loudness of the test frequency is then adjusted to be equal to the loudness at the reference frequency.

[0048] Figure 5 The test panel GUI is shown. Increase the test frequency level by pressing any plus button, or decrease the test frequency level by pressing a minus button, i.e., apply a gain or loss correction. The "Next" and "Previous" buttons jump forward or backward to the next or previous test frequency and / or level.

[0049] The appropriate level range for equal loudness testing is determined based on the measured reference frequency hearing threshold data obtained in step 1. For individuals with normal or moderate hearing loss, 70 phon begins to feel quite uncomfortable, while 60 phon is perceived as high but acceptable. Therefore, testing at levels above 60 phon is generally not recommended. 60 phon is also a very safe level and will not cause any hearing damage. For subjects with an established reference frequency hearing threshold below 40 phon, 60 phon is used as the starting and maximum testing level. For subjects with a hearing threshold of 40 or 50 phon at the reference frequency, 70 phon is set as the starting and maximum level. For subjects with a hearing threshold of 60 phon or higher at the reference frequency, 80 phon is set as the starting and maximum level.

[0050] Preferably, the test frequencies should be 100Hz, 200Hz, 400Hz, 800Hz, 1200Hz, 1600Hz, 2400Hz, 3200Hz, 4800Hz, 6800Hz, 9600Hz, and 12800Hz. Any frequency where excessive hearing loss is present may be discarded. Typically, the higher frequency test points of 9600Hz and 12800Hz are abandoned. The criterion for discarding a test frequency is when the sum of the correction gain levels exceeds the safe limit of sound pressure level or the dynamic limit of the system. When the sum of the correction gain levels reaches 90 dBspl or more, this is considered unsafe and will trigger the cancellation of further testing at that frequency at the current and higher sound levels.

[0051] The equal loudness test begins at the highest sound level (typically 60 phon). The equal loudness test uses at least 10 phon level jumps. If the 60 phon test level exhibits a flat response (i.e., no gain correction is needed at any test frequency), skip 50 phon and set the next test level to 40 phon. Test all frequencies again at the lower levels. If no gain is needed at any test frequency, skip 30 phon and set the next test level to 20 phon. If the 20 phon test level exhibits a flat response, finally test 10 phon. If 20 phon is not flat, test 30 phon as well before finally jumping to 10 phon.

[0052] Generally, if gain correction is required at the current level, the next level should be 10 phons lower than the current level. A 20-phon step size is only used if no gain correction is applied at any test frequency. After gradually decreasing the level by 20 phons, if the response is not flat at the new lower level, the next test level should jump by 10 phons, and then that level should eventually decrease to 10 phons below the current level.

[0053] In the first test series, spanning all test frequencies and sound levels, the correction gain is initially set to 0 dB at each frequency and level, and adjusted up and down from this 0 dB starting point at each frequency and sound level. Following the first test series, selected test frequencies are tested at appropriate sound levels, and the correction gain data is used for a second test run at those selected frequencies and sound levels. The newly acquired correction gain data serves as the starting point for the second test series. For each subsequent test series, the most recently acquired correction gain value is used as the starting point for each frequency and sound level, and this process is repeated for all consecutive test series.

[0054] After the first and second test series, the variability between the first and second results can be evaluated. It has been found that if the correction gain differs significantly from the initial 0 dB, the obtained correction gain value is more likely to be slightly erroneous; therefore, it is important to repeat such test points to check for variability. If high variability is found, the advantage of a point closer to the target starting point is used, and the total variability is calculated. After a suitable maximum number of tests (i.e., 5 to 10), or if the variability between tests no longer increases, a decision is made to stop. The average correction gain among the lowest variability members in the test series is then calculated. A suitable variability threshold is 3 dB.

[0055] Only the first and second test series include all selected test frequencies across all applicable sound levels. The third test series includes only test frequencies at sound levels with variability above a 3dB threshold. Any subsequent test series similarly exclude test frequencies at sound levels with variability below a 3dB threshold, ultimately reaching an overall variability of less than 3dB at all test points or the maximum number of tests limit. The average corrected gain among the lowest variability members of the test series is calculated and used as the final gain correction.

[0056] Step 4, Phon to dBSPL transpose

[0057] Sound pressure level, measured in dBSPL, is the logarithmic unit of the reference level using 20 micropascals of sound pressure; that is, 0 dBSPL equals 20 micropascals of physical sound pressure. The ISO 226 standard provides information on perceived loudness at different sound levels and frequencies. Perceived loudness is measured in phons, which are units based on human perception of loudness. As an example, a person with normal hearing will perceive two sounds at frequencies of 1 kHz and 100 Hz (both at the 30 phon level) as equally loud. However, the 100 Hz sound would require a higher physical sound pressure level than the 1 kHz sound for them to be perceived as equally loud.

[0058] The sound levels for equal loudness testing are based on the ISO 226 standard and measured in Phon meters. Since Phon is a perceived level, the sound level at each frequency test point must be converted from Phon to a technically deriveable physical sound pressure level measured in dBSPL meters.

[0059] The ISO 226 standard contains only a limited number of numerical test points, which are divided into one-third octave bands from 20 Hz to 12500 Hz. The number of numerical test points is also limited.

[0060] The selected frequency points for equal loudness testing do not perfectly coincide with the ISO 226 frequencies, and the granularity of the level data available in the ISO 226 standard is too coarse to be used directly. Therefore, interpolation must be performed between the required data from the numerical test points available in the standard. Many possible mathematical interpolation methods exist, in which case the interpolated value is determined by cubic spline interpolation. Interpolation must be performed on both the level and frequency plane data to achieve the desired granularity.

[0061] During equal loudness testing, at each test frequency and sound level, the sound playback level is first converted from Phon to physical sound pressure level. The sound pressure level is then changed from the initial physical sound pressure level until it is perceived as loud as the reference frequency. The correction gain or loss added to the initial level is measured in logarithmic dB units, not in perceived Phon units.

[0062] As an example, let's consider testing equal loudness at 100Hz and 20Pon. First, the corresponding physical sound pressure level (PSL) at 100Hz must be calculated. The PPL is derived using interpolated ISO 226 data. In this case, 20Pon at 100Hz corresponds to approximately 48.4 dBSPL. Now, suppose the subject perceives the level as too low and requires a 10dB positive correction gain to achieve equal loudness. To achieve equal loudness, the test sound level must be 48.4 dBSPL + 10dB = 58.4 dBSPL. However, 58.4 dBSPL is not equal to 30Pon; the equation 20Pon + 10dB = 30Pon is incorrect. 20Pon at 100Hz = 48.4 dBSPL, while 30Pon at 100Hz = 56.8 dBSPL. The difference of 10Pon at this specific frequency and level is only 8.4 dBSPL, not 10dB. The test sound level is measured in Phon, and the correction gain is measured in dB, which is the basic logarithmic unit, not the perceptual unit.

[0063] In this fourth step, the Phon transpose should be used as a reference for the hearing correction gain. The measured correction gain data is interpolated from the Phon reference to the dBSPL reference. Again, there are many possible mathematical interpolation methods available, in which case the interpolated dBSPL reference value is determined by cubic spline interpolation.

[0064] Any real-time signal processing used for hearing loss correction will receive physical sound pressure (PSP) information from a microphone and generate PSP using a transducer. Of course, in real-time applications, PSP can be converted back and forth to perceptual units. However, this inevitably requires unnecessary additional computational steps, consuming both processing bandwidth and energy. Typically, power supplies in wearable earbuds, hearing aids, or similar products are very limited. Therefore, it is preferable to reference PSP for calculations in real-time signal processing, which constitutes an energy-efficient method to achieve the desired hearing correction. Thus, correction gain data from equal loudness tests at a reference dBSPL level should be available to the real-time signal processing device.

[0065] Step 5: Post-processing of the measured hearing data

[0066] The data post-processing step aims to expand the measured hearing correction data to cover at least the dynamic range between 0 dBSPL and 110 dBSPL. Below 0 dBSPL, the same correction gain as at 0 dBSPL is used, and this level is below the hearing threshold in almost all frequencies. Similarly, levels above 110 dBSPL have the same correction gain as at 110 dBSPL, and these higher levels are not amplified in any case; they are attenuated by the maximum level limiter.

[0067] Equal-loudness hearing tests typically provide data from 10 phon to 60 phon at each frequency in increments of 10 phon. After transposing the phon to dBSPL, data can be obtained by referring to physical dBSPL units. Sometimes a smaller range of levels is available, but at least two, and usually four to six, level measurement points are available.

[0068] First, increase the level of granularity. Interpolation can be implemented mathematically in many ways; in this case, cubic spline interpolation is used between the original data points, followed by multi-order linear-phase FIR averaging filtering on the interpolated dataset. Filtering smooths out local variations in the measured data, thereby improving accuracy. Interpolation can generate as many data points as desired within the available dynamic range, in this case, using an interval of 0.5 dB between the data points.

[0069] Secondly, the dynamic range is extended below the lowest measurement level, down to 0 dBSPL. For this, the lowest-level interpolated and filtered data is used. A linear first-order polynomial derivative is fitted to the derivative of the lower-level portion of the interpolated and filtered data. Then, the data is extended below the lowest measurement level using a first-order polynomial. Similarly, a first-order polynomial is performed on the derivative of the highest-level portion of the interpolated and filtered data using the highest measurement level. The resulting first-order polynomial can then be used to extend the data points above the highest measurement level. Finally, a multi-order linear-phase FIR averaging filter is applied to the entire extended dataset. This filtering smooths the transition between the measured and extended data, improving accuracy. Figure 6 A 3D plot of the measured and extended corrected gain data (referred to as gain sails) is shown at seven different frequencies.

[0070] The next step involves identifying candidate correction frequencies to which correction gain can be applied. Equal loudness tests indicate that frequencies requiring no correction gain will not be used and will be discarded. Frequencies where equal loudness measurements indicate hearing loss is too severe for correction to be applied are also discarded. These indicators suggest that excessive correction gain is needed to restore hearing, or that the test is terminated simply because the sound level is too high. It is not uncommon for the highest frequencies of 9600Hz and 12800Hz to cause problems for people with significant age-related hearing loss, making them unable to hear these frequencies. The remaining candidate frequencies are then further analyzed to find appropriate correction frequencies.

[0071] First, the correction peaks are located within the candidate correction frequencies. In this case, a peak means a frequency where the required correction gain is higher than that of adjacent frequencies. When peaks are identified, a "weight" is calculated for each peak. The weight of a peak is the product of the distance to the next peak and the peak correction gain. Then, the dominant peak with the highest weight is added to the selected frequencies. This process is iterated to find more peaks among the identified peaks, and new peaks are continuously added to the selected frequencies if the distance to an existing peak in the selected frequencies is greater than a threshold. A suitable threshold is twice the number of existing frequencies. When all identified peaks have been investigated, the process stops, and the correction gain frequency is now selected.

[0072] When an appropriate frequency is selected, two optional steps are performed. If the target hearing correction device is known, the correction gain may have been limited in this step to eliminate or at least minimize acoustic feedback problems caused by over-amplification. The maximum correction gain at each frequency, associated with the acoustic feedback characteristics of the hearing correction device, can be beneficially applied to the correction gain data in this stage.

[0073] The final step aims to reduce amplification at the highest sound pressure level. Typically, amplification is not applied at high sound pressure levels because hearing correction does not require it. As an example, Figure 6 The diagram shows the correction gain required to correct for the subject's hearing loss. Although at 20 dBSPL, amplification exceeding 40 dB is required at several frequencies, the correction does not include amplification above approximately 80 dBSPL at any frequency. Figure 6 The required correction gain shown is very common; even for individuals with considerable hearing loss, gain is usually not needed at higher sound pressure levels. In rare cases, when correction gain is present at high sound pressure levels, it is desirable to reduce the gain to avoid additional hearing damage. Therefore, in this step, the correction gain is gradually reduced to zero at sound levels above 90 dBSPL, ensuring that the input level plus gain never produces an output exceeding 100 dBSPL. Two threshold levels, 90 dBSPL and 100 dBSPL, are chosen to mitigate the risk of additional hearing damage and can be adjusted to any other desired level.

[0074] Step 6, Gain Sail Optimization

[0075] Figure 6A 3D plot of gain sail data from individuals facing “cookie-bite” hearing loss is displayed. The gain sail shows the required correction gain (amplification) at frequencies 1 to 7 at sound pressure levels from 0 dBSPL to 120 dBSPL. In this case, the frequencies are 400 Hz, 800 Hz, 1200 Hz, 1600 Hz, 2400 Hz, 3200 Hz, and 12800 Hz. The gain sail reveals the measured correction gain required to fully restore the subject’s hearing and is a collection of correction gain table data for each frequency. Referring to the gain sail, it is clear that no correction gain is needed at higher sound pressure levels, but significant amplification is required at lower sound pressure levels. Above approximately 80 dBSPL, no amplification is needed at any frequency, while at 20 dBSPL, more than 35 dB to 40 dB of amplification is needed in the mid-frequency range to correct the measured hearing loss.

[0076] To restore hearing loss, a real-time signal processing device must dynamically adjust the filter gain at each of seven frequencies based on the input sound pressure level (SPL) at each of those frequencies. In this example, the real-time signal processing device must include seven bandpass filters, which primarily pass each of these frequencies to seven SPL detectors, one detector per frequency. The outputs of the level detectors are then used to calculate the instantaneous required gain for seven peak filters, again one filter per frequency. The number of necessary frequencies varies depending on the situation, and seven in this case is merely an example. In some cases, only one frequency is needed, but in most cases, two to five frequencies are sufficient. Theoretically, all available frequencies might be required, but this is not common.

[0077] Figure 8 The frequency response of eight corrected gain peak filters with appropriate bandwidth is shown. Gain overlap between these filters can be seen in the figure. Significant overlap occurs, especially in the relatively closely spaced mid-frequency range. Close spacing is necessary to obtain correction that matches well with any measured hearing loss. A relatively wide bandwidth of the filters is also necessary; wide-bandwidth filters exhibit good time-domain behavior, while narrow-bandwidth filters produce poor time-domain responses.

[0078] Good temporal behavior is essential, as human hearing is highly sensitive to the temporal characteristics of sound. Whether the sound is produced by striking wood or metal, such as playing the violin or trumpet, it is interpreted by human hearing from the differences in the temporal properties of the sound. Clearly, hearing restoration systems cannot introduce temporal irregularities, as this would degrade sound quality and make hearing and interpreting sound much more difficult. Therefore, corrected gain peak filters must have a wide bandwidth to maintain sound quality, and for this reason, they will always have an overlapping response, such as... Figure 8 exemplified in .

[0079] In order to achieve good tracking between the correction gain and instantaneous sound pressure level in dynamic applications, the correction peak filter bandwidth and the level detector bandpass filter must have a comparable frequency response. Figure 9 The frequency response of the corrected peak filter at 1200 Hz (trace 1) is shown, which overlaps with the appropriate detector bandpass filter at 1200 Hz (trace 2). The bandpass filter cannot be too narrow, as it produces a poor time-domain response and therefore cannot track instantaneous sound pressure levels well. If the time-domain tracking is poor, the measured instantaneous sound pressure level will be incorrect, and the applied correction gain calculated from the measured sound pressure level will also be incorrect. The bandwidth of the bandpass filter cannot be too wide than that of the corrected peak filter, as this will also produce incorrect sound pressure level measurements. In the case of a wideband filter, sounds at frequencies far from the center frequency will be overweighted in the measurement, and therefore the gain at the center frequency will be reduced. This is obviously incorrect, as there is no sound at the center frequency, and the gain applied at the center frequency will become too low. Optimal results are obtained from a sound quality perspective when both filters have similar frequency responses, as shown below. Figure 9 exemplified in .

[0080] Using Figure 8 The filter with the indicated bandwidth is used for correction. Figure 6 The gain shown in the figure indicates that the measured hearing loss, without considering the gain overlap between filters, would result in excessive amplification. Figure 7 The gain flag in the diagram shows the overall amplification without considering gain overlap. It is quite clear that, without considering gain overlap, the total corrected peak filter amplification produces excessive gain, exceeding +90dB at the intermediate frequency (IF), where the gain should only be around +35dB.

[0081] While feedback correction networks can be used to correct overgain in real-time signal processing hearing correction devices, they will always introduce temporal anomalies that degrade sound quality. The severity of the feedback problem becomes particularly apparent when considering a total amplification increase to +90 dB at low input levels (e.g., 20 dB SPL). Then, as an example, when someone begins to speak and silence causes a significant increase in input level, the gain must be immediately reduced to a much lower level; see [reference needed]. Figure 6It might be around 25dB. Then, the gain needs to change by 65dB within a timeframe far less than 10ms, which will cause significant distortion in the sound. However, not only will significant distortion be generated, but the worst problem is gain loss tracking. The gain will initially be too high, and low-level sounds will be unintentionally reproduced at very high levels, resulting in severe initial transient overshoot. Before the gain is turned down, the smooth sound initially sounds like gunshots, which is clearly unsatisfactory. With such a large gain adjustment over such a short time, gain loss tracking will always be present. The feedback will also require additional real-time processing steps, which inevitably consumes processing bandwidth and energy.

[0082] However, a better alternative exists for managing excessive gain, known as gain sail optimization. By using a detector bandpass filter bandwidth that matches the dynamic filter bandwidth, gain table data can be preprocessed and gain sail amplification optimized, thus avoiding significant distortion, gain loss tracking that causes gunshot-like problems, and unnecessary real-time computation. For this, the dynamic filter and detector bandpass filter bandwidths need to be matched so that the detector senses a sound level equal to the gain applied to the dynamic filter. If the filter bandwidths are different, the preprocessed output will be inaccurate.

[0083] Gain sail optimization is a mathematical problem that can be solved using analytical equations or numerical iterative methods. While an analytical solution is theoretically possible, it is completely infeasible due to the complexity and the very large number of variables and equations required. Therefore, numerical solutions are preferred. ELHT utilizes numerical iterative methods to optimize the correction gain data.

[0084] Figure 6 The table data from the equal-loudness hearing test illustrated in the figure are used as the target for the total correction gain at each frequency and sound pressure level involved. Gain optimization aims to remove the excess correction gain accumulated by the correction peak filter boost at adjacent frequencies. Optimization is accomplished through least-squares numerical optimization of the filter gain.

[0085] Once the least squares optimization has found the optimal correction gain for all levels and frequencies, the adjusted value is saved for use by the real-time signal processing device.

[0086] The software functions, digital signal processing, and algorithms can be implemented in various ways, from purely hardware implementations to purely software / firmware implementations or a hybrid of both. The DSP functions in the described invention use code written for a digital signal processor. The described ELHT and algorithms for generating input data for the hearing recovery device are implemented in software running on a personal computer. This software can, of course, be implemented to run on any computing system, such as a phone, tablet, or other device. The software can also be implemented on a specially constructed target system, similar to the audiometer used in the new ELHT system, and cloud computing resources. Detailed Implementation Plan

[0087] Specific embodiments of the present invention are disclosed below. According to the present invention, a software unit intended for use in hearing assessment and / or correction is disclosed, the software unit being configured to perform a method comprising the following steps:

[0088] - Perform reference frequency hearing tests to establish reference frequencies for each ear of the test subject; and

[0089] - Perform equal-loudness hearing tests on each ear.

[0090] The steps of performing a reference frequency hearing test involve: starting from a first frequency located within a given frequency range, providing sound to the test subject at a set sound level to obtain or not obtain a hearing response, wherein the set sound level is selected as the lowest test sound level within the range of -10 phon to 50 phon; changing the sound level at the first frequency; if the test subject provides a hearing response at the selected lowest test sound level within the range of -10 phon to 50 phon, then selecting the first frequency as the reference frequency; otherwise, selecting another frequency within the given frequency range; and repeating the iterative process with the new frequency within the given frequency range until the software unit detects a frequency X for which the test subject provides a hearing response at the selected lowest test sound level within the range of -10 phon to 50 phon, or the frequency is a frequency that has the lowest sound level hearing response from the test subject for the set frequency to be tested, and then selecting the frequency X as the reference frequency.

[0091] According to a particular implementation, the selected minimum test sound level is chosen in the range of 0 phon to 20 phon, preferably in the range of 5 phon to 15 phon. As a suitable example, the selected minimum test sound level is then at or near 10 phon.

[0092] According to one specific implementation plan, the step of changing the sound level is performed by gradually reducing the sound level at the first frequency until no hearing response is obtained from the test subject.

[0093] Furthermore, according to another embodiment, the software unit is programmed to select a frequency closer to the middle of the given frequency range rather than a frequency further out, preferably, when both are at equal distances from the middle of the given frequency range, a lower frequency is selected rather than a higher frequency.

[0094] Furthermore, according to another implementation, the software unit is programmed to select new frequencies in iteration according to priority order. This can be done through a set and implemented priority list.

[0095] Furthermore, according to another embodiment, the process of changing the sound level is performed in steps of at least 1 phon, preferably at least 5 phons, and more preferably at least 10 phons, until an absolute value of 10 phons is reached or the test subject provides no hearing response. According to one embodiment, the sound level is changed by gradually decreasing it in steps of at least 1 phon, preferably at least 5 phons, and more preferably at least 10 phons, until an absolute value of 10 phons is reached or the test subject provides no hearing response.

[0096] Furthermore, according to one embodiment, the software unit is configured to perform a left-right channel level balance test, which includes balancing the perceived loudness of the reference frequency between the left and right ears of the test subject.

[0097] According to one implementation, the step of balancing the perceived loudness of the reference frequency between the left and right ears of the test subject is performed before the equal loudness hearing test for each ear.

[0098] According to another embodiment, the software unit is configured to perform the left-right channel level balance test between the step of performing the reference frequency hearing test and the step of performing the equal loudness hearing test. Furthermore, according to one embodiment, the software unit is configured to perform the left-right channel level balance test at all set measurement sound levels. Furthermore, according to yet another embodiment, the step of performing the equal loudness hearing test for each ear is performed by comparing the perceived loudness at the reference frequency for each ear with a plurality of test frequencies, one frequency per ear.

[0099] Furthermore, the software unit can be configured to perform the equal-loudness hearing test by providing different frequencies other than a set reference frequency and comparing the responses from the test subject at multiple appropriate sound pressure levels above the test subject's hearing threshold. According to another embodiment, the software unit is configured to perform a first series of equal-loudness hearing tests by providing the test subject with a sequence of different sound levels and testing whether gain correction is needed at each sound level. Furthermore, according to yet another embodiment, the sequence of different sound levels in the first test series is provided by progressively decreasing the sound levels until a sound level requiring gain correction is found, preferably wherein the subsequent sound level being tested is higher than the sound level at which gain correction is established.

[0100] According to one embodiment, the software unit is configured to perform a second test series, wherein established correction gain values ​​from the first test series are used to provide different sound levels of the test subject to obtain new correction gain values, preferably wherein the established correction gain values ​​are iteratively used in the next test series. Furthermore, the software unit may be configured to evaluate the variability of the correction gain values ​​between different test series. Additionally, when further testing for a particular sound level is stopped, a maximum threshold of this variability can be used as input for a decision; preferably, the maximum threshold is 3 dB. According to yet another embodiment, the software unit is configured to calculate the average correction gain among the lowest variable members in the performed test series and set this average correction gain as the final gain correction. Furthermore, the software unit is suitably configured to perform the equal-loudness hearing test using vibrato, preferably using a vibrato modulation range of 1% to 10%, preferably at a modulation frequency in the range of 5 Hz to 30 Hz.

[0101] As should be clear from the above, according to the present invention, the software unit is preferably also configured to collect data obtained from the test subject. Furthermore, according to another embodiment, the software unit is configured to preferably discard the one or more test frequencies based on the test subject's hearing loss at one or more test frequencies.

[0102] Furthermore, according to one embodiment of the invention, the software unit is arranged to appropriately set the spl range for the equal-loudness hearing test within a range of 10 phon to 70 phon, wherein the equal-loudness hearing test is performed by changing the sound level, preferably in steps of at least 1 phon, more preferably in steps of at least 5 phon, and even more preferably in steps of at least 10 phon. Furthermore, according to one embodiment, the step of setting the spl range for the equal-loudness hearing test is performed by gradually decreasing the value from the high end of the set range.

[0103] According to one implementation, the software unit is located in a computer unit or a mobile device such as a telephone or tablet, or is located as part of an embedded system, such as in a specially constructed product.

[0104] Furthermore, according to another embodiment, the software unit or connected data calculation software is arranged to perform a transpose of the obtained data from Phon to absolute dBSpl values ​​by interpolating the measured correction gain data from a Phon reference to a dBSpl reference. Relatedly, and as should be noted below, according to the invention, any type of setup, more or less, with different connected software units is possible. According to one embodiment of the invention, the software unit or connected data calculation software is arranged to preferably extend the dynamic range below the lowest measurement level and / or above the highest measurement level by using a first-order polynomial, more preferably by multi-order linear-phase FIR averaging filtering. In this regard, it should be noted that the application can be provided in a telephone or computer unit and can then be connected to another program, such as one performing additional calculations in the cloud.

[0105] Furthermore, according to one embodiment, the software unit or connected data calculation software is configured to identify and select candidate correction frequencies to which the correction gain can be applied by finding peak frequencies where the required correction gain is higher than that of adjacent frequencies, preferably then to calculate the weight of each identified peak based on the distance to a nearby peak and the correction gain of a particular peak, more preferably then to identify the dominant peak with the highest weight and add the dominant peak to the selected frequency.

[0106] Furthermore, according to another embodiment, the software unit or connected data calculation software is arranged to perform gain sail optimization, which involves compensating for the gain contribution of adjacent filters in the presence of aggregated amplification from adjacent filters. According to the invention, this gain sail operation can be performed in different locations, such as in an additional software unit or in the main software. Furthermore, according to one embodiment, the gain sail optimization involves preferably removing the excess correction gain accumulated by the boosting of adjacent frequency correction peak filters by performing least-squares numerical optimization of the filter gain.

[0107] As implied above, the software unit according to the invention can be part of a system. Accordingly, according to one embodiment, a system is provided comprising the software unit according to the invention, wherein the software unit includes a digital signal processing (DSP) unit, or is connected to another software unit or device including a digital signal processing (DSP) unit, said DSP unit being arranged to process and compensate for data obtained from the test object.

[0108] According to one embodiment, the digital signal processing (DSP) unit involves one or more dynamic filters for compensating the data, preferably multiple dynamic filters for compensating the data, preferably two to twenty dynamic filters for compensating the data, preferably each dynamic filter operates with a dynamically varying gain that depends on the dynamically varying level of the input signal at the filter frequency. Furthermore, according to yet another embodiment, the DSP unit is arranged to digitally process amplified data and filter center frequencies, preferably by involving one or more filter blocks. Additionally, according to one embodiment, the DSP unit involves one or more dynamic filters for compensating the data, wherein each dynamic filter involved has a center frequency in the range of 100 Hz to 12.8 kHz.

[0109] According to another embodiment, the one or more dynamic filters operate with a dynamically varying gain, which depends on the dynamically varying input signal level at a specific filter frequency. Furthermore, suitably, the one or more filter blocks involved in the digital signal processing (DSP) unit include a bandpass filter, a sound pressure level detector, and a dynamic filter. Furthermore, according to another embodiment, the bandpass filter is arranged to filter out the signal level at the dynamic filter frequency, and the detector is arranged to measure the signal level at the dynamic filter frequency and suppress sound signals present at other frequencies.

[0110] Furthermore, according to one embodiment, the bandpass filter is low-order, preferably second-order, and more preferably Q is less than 1. Furthermore, according to yet another embodiment, the bandpass filter and the dynamic filter have matched bandwidths, and the dynamic filter is also considered a peak filter.

[0111] Furthermore, according to one embodiment, the one or more filter blocks also include a gain table that uses, for example, table data from an algorithm and converts the current input level into a gain setting in the dynamic filter.

[0112] Furthermore, according to another embodiment, the amplified data and filter center frequency are imported from the software algorithm into the digital signal processing (DSP) unit, and the DSP unit includes one or more filter blocks that provide dynamically varying amplification, each filter block handling a separate frequency range.

[0113] Furthermore, according to another implementation, multiple filter blocks are used in the digital signal processing (DSP) unit.

Claims

1. A software unit intended for use in hearing assessment, the software unit being configured to perform a method comprising the steps of: - Perform a reference frequency hearing test to establish a reference frequency for each ear of the test subject; as well as - Perform equal-loudness hearing tests on each ear. The steps of performing the reference frequency hearing test involve: starting from a first frequency located within a given frequency range, providing sound to the test subject at a set sound level to obtain or not obtain a hearing response, wherein the set sound level is selected as the lowest test sound level in the range of -10 phon to 50 phon; and changing the sound level at the first frequency. If the test subject provides a hearing response at the selected lowest test sound level within the range of -10 phon to 50 phon, then the first frequency is selected as the reference frequency; otherwise, another frequency within the given frequency range is selected. The iterative process is repeated with new frequencies within the given frequency range until the software unit detects a frequency X for which the test subject provides a hearing response at the selected lowest test sound level within the range of -10 phon to 50 phon, or the frequency is a frequency that has the lowest sound level hearing response from the test subject for a set frequency to be tested, and then the frequency X is selected as the reference frequency.

2. The software unit according to claim 1, wherein the selected minimum test sound level is selected in the range of 0 phon to 20 phon, preferably in the range of 5 phon to 15 phon.

3. The software unit according to claim 1 or 2, wherein the step of changing the sound level is performed by gradually decreasing the sound level at the first frequency until no hearing response is obtained from the test subject.

4. The software unit according to any one of claims 1 to 3, wherein the software unit is programmed to select a frequency closer to the middle of the given frequency range rather than a frequency further out, preferably, when both are at equal distances from the middle of the given frequency range, also selecting a lower frequency rather than a higher frequency.

5. The software unit according to any one of claims 1 to 4, wherein the software unit is programmed to select a new frequency in priority order during iteration.

6. The software unit according to any one of claims 1 to 5, wherein the process of changing the sound level is performed in steps of at least 1 phon, preferably at least 5 phon, more preferably at least 10 phon, until an absolute value of 10 phon is reached or the test subject does not provide a hearing response.

7. The software unit of claim 6, wherein changing the sound level is performed by gradually decreasing the sound level in steps of at least 1 phon, preferably at least 5 phon, more preferably at least 10 phon, until an absolute value of 10 phon is reached or the test subject does not provide a hearing response.

8. The software unit according to any one of claims 1 to 7, wherein the software unit is arranged to perform a left-right channel level balance test, the left-right channel level balance test comprising balancing the perceived loudness of the reference frequency between the left and right ears of the test subject.

9. The software unit of claim 8, wherein the step of balancing the perceived loudness of the reference frequency between the left and right ears of the test subject is performed prior to the equal-loudness hearing test for each ear.

10. The software unit of claim 8 or 9, wherein the software unit is arranged to perform the left and right channel level balance test between the step of performing the reference frequency hearing test and the step of performing the equal loudness hearing test.

11. The software unit according to any one of claims 8 to 10, wherein the software unit is configured to perform a left and right channel level balance test at all set measurement sound levels.

12. The software unit according to any one of claims 1 to 11, wherein the step of performing an equal loudness hearing test for each ear is performed by comparing the perceived loudness at the reference frequency for each ear with a plurality of test frequencies, one frequency per ear.

13. The software unit according to any one of claims 1 to 12, wherein the software unit is arranged to perform the equal loudness hearing test by providing different frequencies other than a set reference frequency and comparing responses from the test subject at a plurality of appropriate sound pressure levels above the hearing threshold of the test subject.

14. The software unit according to any one of claims 1 to 13, wherein the software unit is arranged to perform a first series of tests of equal loudness hearing by providing the test subject with a sequence of different sound levels and testing whether gain correction is required at each sound level.

15. The software unit of claim 14, wherein the sequence of the different sound levels in the first test series is provided by progressively decreasing the sound levels until a sound level is found that requires gain correction, preferably wherein the subsequent sound level being tested is higher than the sound level at which the gain correction is established.

16. The software unit of claim 14 or 15, wherein the software unit is arranged to perform a second test series, wherein the established correction gain value from the first test series is used to provide different sound levels of the test object to obtain a new correction gain value, preferably wherein the established correction gain value is iteratively used for the next test series.

17. The software unit of claim 16, wherein the software unit is arranged to evaluate the variability of the correction gain value between different test series.

18. The software unit of claim 17, wherein when further testing for a particular sound level is stopped, the maximum threshold of the variability is used as an input for decision-making, preferably, the maximum threshold is 3 dB.

19. The software unit of claim 17 or 18, wherein the software unit is arranged to calculate the average correction gain among the lowest variable members in the performed test series and set the average correction gain as the final gain correction.

20. The software unit according to any one of claims 1 to 19, wherein the software unit is arranged to perform the equal loudness hearing test by using a vibrato, preferably by using a vibrato modulation range of 1% to 10%, preferably at a modulation frequency in the range of 5 Hz to 30 Hz.

21. The software unit according to any one of claims 1 to 20, wherein the software unit is arranged to collect data obtained from the test object.

22. The software unit according to any one of claims 1 to 21, wherein the software unit is arranged to preferably discard the one or more test frequencies based on the hearing loss of the test subject at one or more test frequencies.

23. The software unit according to any one of claims 1 to 22, wherein the software unit is arranged to appropriately set the spl range for the equal loudness hearing test in the range of 10 phon to 70 phon, and wherein the equal loudness hearing test is performed by changing the sound level, preferably at least 1 phon per step, more preferably at least 5 phon per step, and even more preferably at least 10 phon per step.

24. The software unit of claim 23, wherein the step of setting the spl range for the equal loudness hearing test is performed by gradually decreasing from a high value within the set range.

25. The software unit according to any one of claims 1 to 24, wherein the software unit is arranged in a computer unit or a mobile device such as a telephone or tablet computer, or is arranged as part of an embedded system, such as in a specially constructed product.

26. The software unit according to any one of claims 1 to 25, wherein the software unit or the connected data calculation software is arranged to perform a transpose of the obtained data from Phon to absolute dBSpl values ​​by interpolating the measured correction gain data from a Phon reference to a dBSpl reference.

27. The software unit of claim 26, wherein the software unit or the connected data calculation software is arranged to preferably extend the dynamic range below the lowest measurement level and / or above the highest measurement level by using a first-order polynomial, more preferably by using a multi-order linear-phase FIR averaging filter.

28. The software unit of claim 26 or 27, wherein the software unit or connected data calculation software is configured to identify and select candidate correction frequencies to which the correction gain can be applied by finding peak frequencies where the required correction gain is higher than that of adjacent frequencies, preferably then to calculate a weight for each identified peak based on the distance to a nearby peak and the correction gain of a particular peak, more preferably then to identify a master peak with the highest weight and add the master peak to the selected frequency.

29. The software unit according to any one of claims 26 to 28, wherein the software unit or connected data computation software is arranged to perform gain sail optimization, the gain sail optimization involving compensation for the gain contribution of adjacent filters in the presence of aggregation of amplification from adjacent filters.

30. The software unit of claim 29, wherein the gain optimization involves preferably removing excess correction gain accumulated by the adjacent frequency correction peak filter boost by performing least-squares numerical optimization of the filter gain.

31. A system comprising a software unit according to any one of claims 1 to 30, wherein the software unit comprises a digital signal processing (DSP) unit, or is connected to another software unit or device comprising a digital signal processing (DSP) unit, the digital signal processing (DSP) unit being arranged to process and compensate for data obtained from the test object.

32. The system of claim 31, wherein the digital signal processing (DSP) unit relates to one or more dynamic filters for compensating the data, preferably to a plurality of dynamic filters for compensating the data, preferably to two to twenty dynamic filters for compensating the data, preferably, each dynamic filter operates with a dynamically varying gain, the gain depending on the dynamically varying level of the input signal at the filter frequency.

33. The system of claim 31 or 32, wherein the digital signal processing (DSP) unit is arranged to perform digital signal processing on the amplified data and the filter center frequency, preferably by involving one or more filter blocks.

34. The system of claim 32 or 33, wherein the digital signal processing (DSP) unit relates to one or more dynamic filters for compensating the data, and wherein each dynamic filter relates to has a center frequency in the range of 100 Hz to 12.8 kHz.

35. The system according to any one of claims 32 to 34, wherein the one or more dynamic filters operate with dynamically varying gain, the gain depending on the dynamically varying level of the input signal at a particular filter frequency.

36. The system according to any one of claims 33 to 35, wherein the one or more filter blocks involved in the digital signal processing (DSP) unit include a bandpass filter, a sound pressure detector, and a dynamic filter.

37. The system of claim 36, wherein the bandpass filter is arranged to filter out the signal level at the frequency of the dynamic filter and the detector is arranged to measure the signal level at the frequency of the dynamic filter and suppress sound signals present at other frequencies.

38. The system according to claim 36 or 37, wherein the bandpass filter is low-order, preferably second-order, and more preferably Q is less than 1.

39. The system according to any one of claims 36 to 38, wherein the bandpass filter and the dynamic filter have matched bandwidths.

40. The system of any one of claims 33 to 39, wherein the one or more filter blocks further include a gain table that uses, for example, table data from an algorithm and converts the current input level into a gain setting in the dynamic filter.

41. The system according to any one of claims 33 to 40, wherein the amplified data and the filter center frequency are imported from the software algorithm into the digital signal processing (DSP) unit, and wherein the DSP unit includes one or more filter blocks that provide dynamically varying amplification, each filter block handling a separate frequency range.

42. The system according to any one of claims 33 to 41, wherein a plurality of filter blocks are used in the digital signal processing (DSP) unit.

Images