Beta Test: Earful -- a hearing test App

[Dreyfus]

I just did three runs with my minimum hearing threshold for the ATH-R70X.
Those are the results I got for pure sine, white noise and pink noise (Q 4.3):

The results do line up well with my expectations described in the Peace Calibration Tool Thread.
I will continue to use 1/3 octave pink noise as suggested by David Griesinger since it offers a smooth and non-fatiguing procedure.

Technically I had some issue with the WASAPI exclusive mode for my RME Babyface Pro which added some distortion when boosting tones below -80 dBFS. Possibly a conflict with my driver setup + the EQ APO integration running in the background. I then switched back to the normal WASAPI mode: all fine!

One thing I would really love to see is a hotkey feature for quickly changing the bands and their level. I would suggest to bind them to the arrow keys. That would allow the user to tune the curve without having to look at the screen. Would be better for longer concentration and could also prevent bias to tune the curve to a specific target (confirm the known F-M or just the latest preference curve the user adjusted, for example).

Regards,
Dreyfus

 

[jae]

Possible to have an option to have a separate graphed line for R/L as well?

 

[pkane]

Possible to have an option to have a separate graphed line for R/L as well?

You can save one, then load it as Data Set 2 to overlay over the other.

 

[Muflatschki Gurkovski]

Hi, I just signed in to ask you for a favour, because I want to use earful for a special purpose.
My aim is similiar to the one of griesingers dgsonicfocus, where you match the equal loudness of different frequency noise bands and finally equalize your headphones in that way they make alien binaural recordings sound more in front of you.
It's feasible because equal loudness curves regard the personal indivdual cues of your earcanal.
BUT I really struggle with comparing the perceptive loudness of different noise bands to each other caused by the accompanied uncertainty of deciding which loudness matches the other best.
So I pursue another approach which could make the procedure much easier but it is only feasible if earful provides a special option.
I want to match the loudness of noise bands from my headphone directly to the same from a (flattened) speaker, so in the end it's just the loudness to compare and no different noise bands to each other.
Of course that requires a headphone with very low dampening like open back type which I use for this case.
For maximum ease the signal from the headphone must switch to the speaker output without any interruption. It has to sound like ONE seamless sound because the human "acoustic memory" (and also mine) is very bad at all.
The problem is that earful sets a new signal sound start intervall as soon as I adjust the loudness or alter the frequency band, so may you add an option to set the intervall start point only once and keep the intervall fix from then on? Or would that take too much effort?
I would open earful twice and choose the speaker output for one of them and the headphone output (e.g. the external dac interface) for the second one which definitly worked for me.
Then I will synchronize the two intervall starts until the sound of speaker and headphone alternate seamlessly and with that method I can go on to repeat it for all the freqency bands.
It would work like an abbreviation of dgsonicfocus' method with the caveat of the need for less dampening headphones.
I also plan to measure the dampening of my headphone (for different frquencies) to compensate the eventual loudness drop by amplifying the speaker magnitude appropriately.
I can not use dgsonicfocus with my intended method because it doesn't allow you to choose a noise band without alternating to the 500hz frequency band + also sets a new intervall start with the adjustment of loudness or noise band + it needs asio to work, so I cannot dedicate the signals to different outputs I think.
Hence it would be very helpful for achieving my decribed goal and l will appreciate for your help, hope it is not beyond your possibilities.

 

[pkane]

Hi @Muflatschki Gurkovski ,

Interesting suggestion. Let me ask you this: would it work for your purpose if you could have one instance of Earful outputting alternating tones/noise bands into different sound devices? Obviously, with an ability to adjust the volume independently, so one of them (speakers, for example) can be the reference, while the second one can be the headphones for measuring?

Hi, I just signed in to ask you for a favour, because I want to use earful for a special purpose.
My aim is similiar to the one of griesingers dgsonicfocus, where you match the equal loudness of different frequency noise bands and finally equalize your headphones in that way they let alien binaural recordings sound more in front of you.
It's feasible because equal loudness curves regard the personal indivdual cues of your earcanal.
BUT I really struggle with comparing the perceptive loudness of different noise bands to each other caused by the accompanied uncertainty of deciding which loudness matches the other best.
So I pursue another approach which could make the procedure much easier but it is only feasible if earful provides a special option.
I want to match the loudness of noise bands from my headphone directly to the same from a (flattened) speaker, so in the end it's just the loudness to compare and no different noise bands to each other.
Of course that requires a headphone with very low dampening like open back type which I use for this case.
For maximum ease the signal from the headphone must switch to the speaker output without any interruption. It has to sound like ONE seamless sound because the human "acoustic memory" (and also mine) is very bad at all.
The problem is that earful sets a new signal sound start intervall as soon as I adjust the loudness or alter the frequency band, so may you add an option to set the intervall start point only once and keep the intervall fix from then on? Or would that take too much effort?
I would open earful twice and choose the speaker output for one of them and the headphone output (e.g. the external dac interface) for the second one which definitly worked for me.
Then I will synchronize the two intervall starts until the sound of speaker and headphone alternate seamlessly and with that method I can go on to repeat it for all the freqency bands.
It would work like an abbreviation of dgsonicfocus' method with the caveat of the need for less dampening headphones.
I also plan to measure the dampening of my headphone (for different frquencies) to compensate the eventual loudness drop by amplifying the speaker magnitude appropriately.
I can not use dgsonicfocus with my intended method because it doesn't allow you to choose a noise band without alternating to the 500hz frequency band + also sets a new intervall start with the adjustment of loudness or noise band + it needs asio to work, so I cannot dedicate the signals to different outputs I think.
Hence it would be very helpful for achieving my decribed goal and l will appreciate for your help, hope it is not beyond your possibilities.

 

[Muflatschki Gurkovski]

...would it work for your purpose if you could have one instance of Earful outputting alternating tones/noise bands into different sound devices? Obviously, with an ability to adjust the volume independently, so one of them (speakers, for example) can be the reference, while the second one can be the headphones for measuring?

Hmm, mind, that the signal must switch from one source (headphone) to the other (speaker) seamlessly and

Changing the frequency band
1. must happen for both signals simultaneously (so one click for both)
2.must restart the interval for both signals simultaneously (or keeps them alternating like before), so no false shifting/ offset occurs

3.adjusting the volume for one device (headphone) must NOT reset the intervall start or at least keeps the signals alternating like before.

Because all of that I have thought my idea leads to less programming work for you but on the other side yours is much more comfortable while mine would be so awkward.

If the seamless alternating between the two signals is not a big problem for you I suggest to realise the rest by implementing a "couple function" (often marked as a little chain)
In my case I need to couple the (same) frequency bands of both signals and the volume remains independant.
But then implementing the volume adjustment without coherent reset of the interval of only this signal remains problematic, I assume?

An option to apply one curve dedicated to one device would also be very useful for me because I have to inject the eq for the speaker flattening.
I could use eq apo for this, but it doesn't work with asio, which my interface needs.
Otherwise I am forced to use my notebook headphone jack for the speaker connection, which is not a real line-out. (Ok, or perhaps bluetooth as an alternative option)

Thank You very much for your reply!

 

[pkane]

I have to admit I've not done a whole lot of testing on this new version, so hoping that you can help me

Earful v1.0.10:
https://distortaudio.org/earful.html

This version lets you chose from three types of testing:

• Threshold
• Equal Loudness
• EQ Two Devices

Threshold is the threshold of hearing test, works just like it did previously to measure the lowest level you can still hear at each frequency.

Equal Loudness will alternate between two tones, 1kHz reference tone at a specific level (set by Start Vol setting) and the desired frequency test tone. You adjust the level of the test tone until it sounds equal in level to the 1kHz tone. "REFERENCE" badge will flash when the 1kHz reference tone is playing. This test is meant to help compute your own personal equal-loudness curves similar to Fletcher-Munson. If you use the SPL calibration feature, you can even compute equal loudness curves at the specific phon levels to match Fletcher-Munson or ISO 226:2003 curves (phon is SPL level in dB at 1kHz) .

EQ Two Devices will alternately play reference and test tones using two different audio devices. The reference can be your speakers, while the test device can be headphones, for example. PLEASE NOTE: both devices should be set to the same sampling rate. Reference tone will change frequency as you move around the curve, but the reference volume will remain the same. This time, the volume is set by the Reference Volume setting in the two-device selection window. Your job is to adjust the level that you hear through the test device (headphones) to equal in loudness to the sound you hear from your speakers. "REFERENCE" badge will flash when the reference tone is playing.

PLEASE NOTE: both devices should be set to the same sampling rate.

To bring up the audio device selector again, just re-select EQ Two Devices from the Test Type drop-down list.

I'm sure there are still some rough edges, but let's hear if this is at least heading in the right direction!

Hmm, mind, that the signal must switch from one source (headphone) to the other (speaker) seamlessly and

Changing the frequency band
1. must happen for both signals simultaneously (so one click for both)
2.must restart the interval for both signals simultaneously (or keeps them alternating like before), so no false shifting/ offset occurs

3.adjusting the volume for one device (headphone) must NOT reset the intervall start or at least keeps the signals alternating like before.

Because all of that I have thought my idea leads to less programming work for you but on the other side yours is much more comfortable while mine would be so awkward.

If the seamless alternating between the two signals is not a big problem for you I suggest to realise the rest by implementing a "couple function" (often marked as a little chain)
In my case I need to couple the (same) frequency bands of both signals and the volume remains independant.
But then implementing the volume adjustment without coherent reset of the interval of only this signal remains problematic, I assume?

An option to apply one curve dedicated to one device would also be very useful for me because I have to inject the eq for the speaker flattening.
I could use eq apo for this, but it doesn't work with asio, which my interface needs.
Otherwise I am forced to use my notebook headphone jack for the speaker connection, which is not a real line-out. (Ok, or perhaps bluetooth as an alternative option)

Thank You very much for your reply!

 

[Muflatschki Gurkovski]

I have to admit I've not done a whole lot of testing on this new version, so hoping that you can help me

Earful v1.0.10:
https://distortaudio.org/earful.html

This version lets you chose from three types of testing:

• Threshold
• Equal Loudness
• EQ Two Devices

View attachment 101885​

Threshold is the threshold of hearing test, works just like it did previously to measure the lowest level you can still hear at each frequency.

Equal Loudness will alternate between two tones, 1kHz reference tone at a specific level (set by Start Vol setting) and the desired frequency test tone. You adjust the level of the test tone until it sounds equal in level to the 1kHz tone. "REFERENCE" badge will flash when the 1kHz reference tone is playing. This test is meant to help compute your own personal equal-loudness curves similar to Fletcher-Munson. If you use the SPL calibration feature, you can even compute equal loudness curves at the specific phon levels to match Fletcher-Munson or ISO 226:2003 curves (phon is SPL level in dB at 1kHz) .

EQ Two Devices will alternately play reference and test tones using two different audio devices. The reference can be your speakers, while the test device can be headphones, for example. PLEASE NOTE: both devices should be set to the same sampling rate. Reference tone will change frequency as you move around the curve, but the reference volume will remain the same. This time, the volume is set by the Reference Volume setting in the two-device selection window. Your job is to adjust the level that you hear through the test device (headphones) to equal in loudness to the sound you hear from your speakers. "REFERENCE" badge will flash when the reference tone is playing.

PLEASE NOTE: both devices should be set to the same sampling rate.

To bring up the audio device selector again, just re-select EQ Two Devices from the Test Type drop-down list.

View attachment 101886

I'm sure there are still some rough edges, but let's hear if this is at least heading in the right direction!

I have tested it for a short time and want to give some feedback.
It worked as requested after a few tries with one minor flaw, which nevertheless is very crucial for me.
The two signals overlap for a very short time and that leads to a short increase of loudness during the transition, which confuses the brain and interrupts the process of comparing their loudnesses.
Perhaps it depends on the used devices and you can't ensure the perfect transition.
I suggest to provide a user setting for the right delay or to correct it as much as possible.

Another minor flaw I noticed is the selected channel (left/right/both) always applies to both devices, which isn't useful in all cases, although it's mostly suitable.

For anybody, who wants to proceed like me for matching different loudnesses of the same noise band from headphone and speaker, bear the following in mind:
The perceived total loudness is composed of the loudness of the left and the right ear, so listening to a single speaker regardless of mid or left/right position will always affect both ears.
For that reason comparing the loudness of one headphone channel (left or right) with the one from a single speaker using BOTH ears doesn't make sense.
I infer, it's indicated to seal the unnecessary ear during the test, but even then the signal will be heard (at lower loudness) by that ear.

After the loudness equing is done for the left and right ear (separately), I would proof the result by enabling both channels (left and right) with applied headphone eqs (also for both channels at once of course) and comparing the noise bands loudness with the speaker loudness.
Therefor a two channel eq is needed with both channel eqs applicable at the same time.

 

[pkane]

I have tested it for a short time and want to give some feedback.
It worked as requested after a few tries with one minor flaw, which nevertheless is very crucial for me.
The two signals overlap for a very short time and that leads to a short increase of loudness during the transition, which confuses the brain and interrupts the process of comparing their loudnesses.
Perhaps it depends on the used devices and you can't ensure the perfect transition.
I suggest to provide a user setting for the right delay or to correct it as much as possible.

Another minor flaw I noticed is the selected channel (left/right/both) always applies to both devices, which isn't useful in all cases, although it's mostly suitable.

For anybody, who wants to proceed like me for matching different loudnesses of the same noise band from headphone and speaker, bear the following in mind:
The perceived total loudness is composed of the loudness of the left and the right ear, so listening to a single speaker regardless of mid or left/right position will always affect both ears.
For that reason comparing the loudness of one headphone channel (left or right) with the one from a single speaker using BOTH ears doesn't make sense.
I infer, it's indicated to seal the unnecessary ear during the test, but even then the signal will be heard (at lower loudness) by that ear.

After the loudness equing is done for the left and right ear (separately), I would proof the result by enabling both channels (left and right) with applied headphone eqs (also for both channels at once of course) and comparing the noise bands loudness with the speaker loudness.
Therefor a two channel eq is needed with both channel eqs applicable at the same time.

Thanks for testing. Yes, there's a little overlap between the two signals as there is a need to transition slowly from one to the other (about a 1/4 of a second or so) so as to not result in sudden pops and clicks. I'll see if I can adjust the timing so that they overlap less, but that may leave a short pause between the two.

 

[Muflatschki Gurkovski]

A short pause would cause a minor problem than an increasing volume because the following true signal appears always lower then, what is irritating for the brain during the comparison.
And if you let the user set the right timing himself or is that to difficult to intergrate?

 

[pkane]

A short pause would cause a minor problem than an increasing volume because the following true signal appears always lower then, what is irritating for the brain during the comparison.
And if you let the user set the right timing himself or is that to difficult to intergrate?

I'll have to take a look. Timing with fade-in/fade-out was integrated into the signal generator. Since there are now two devices and two signal generators, it may take some work to make the timing between them configurable

 

[Muflatschki Gurkovski]

Hi, a short question:
Does earful provide an option to replay all the frequencies with the equed loudness in one run (without alternating to reference band)?
That's, what I am also missing from Griesingers dgsonic focus, because it's a nice way to proof your results.
When played the equed bands in a row, variances are rather noticeable and a quick adjustment can be made.
The repeated refer to a reference band let the brain get tired after a while.
The "manloud eq" of the realiser a 16 works like that and I appreciate this approach.

 

[pkane]

Hi, a short question:
Does earful provide an option to replay all the frequencies with the equed loudness in one run (without alternating to reference band)?
That's, what I am also missing from Griesingers dgsonic focus, because it's a nice way to proof your results.
When played the equed bands in a row, variances are rather noticeable and a quick adjustment can be made.
The repeated refer to a reference band let the brain get tired after a while.
The "manloud eq" of the realiser a 16 works like that and I appreciate this approach.

If you select "Threshold" test type and then move through the points on the curve with left/right arrows, you'll hear only the corrected levels at each of the frequencies. There is a pause between tones, though.

 

[pkane]

Earful v1.0.11 posted.

Changes in 1.0.11

• Changed: timing in EQ Two Devices to reduce overlap and interference between the reference and test signals
• Changed: Reference device in EQ Two Devices test now always plays in both channels, Left/Right/Both control only applies to the test device

 

[Muflatschki Gurkovski]

I tested it, albeit the overlap is reduced, it still reveals a rise of loudness in transition especially apparent with sinussweep/ warb. It' s too distracting, so I recommend rather a short pause than a minimal overlap.
Aside from that I found out, that the two devices loudness method works best with the clear sinesweep.

Thanks for your efforts, I wish you a happy new year!

 

[jae]

I did not try the new version yet and will do so later but I noticed a bug in 1.0.9. If I changed tone type (warble, single, noise etc.) while playing a tone, sound would come through the left channel even though only the right was selected. Not sure if that bug carries over

Would also be great if you could add a constant tone and perhaps square wave/sawtooth/triangle etc.

 

[jae]

Noticed in the latest version, I can no longer pin it to my taskbar because of the warning screen thing

 

[hutt132]

I actually tried equal loudless like this a while ago, but couldn't wrap my head around how to apply it to my headphone's FR, given that they themselves are not flat.

Using HD600, starting volume -20:

20, -20
24, -3.5
31, -5.5
39, -8
48, -9
60, -11.5
76, -14
95, -15.5
118, -17.5
148, -20.5
185, -20.5
232, -21.5
289, -21.5
362, -21.5
452, -21.5
565, -22
706, -21.5
883, -20.5
1103, -21
1379, -21
1723, -21.5
2154, -21.5
2691, -22
3363, -22
4203, -22
5252, -22
6563, -22.5
8202, -17.5
10249, -27.5
12808, -21.5
16005, -18.5
19999, -20

 

[pkane]

I did not try the new version yet and will do so later but I noticed a bug in 1.0.9. If I changed tone type (warble, single, noise etc.) while playing a tone, sound would come through the left channel even though only the right was selected. Not sure if that bug carries over

Would also be great if you could add a constant tone and perhaps square wave/sawtooth/triangle etc.

Yes, that's a bug. I think for now the easiest thing is to press pause and then play to get everything back to normal. It appears the dynamic change of tone type while playing isn't setting everything up right. I'll see if I can fix that in the next version.

Noticed in the latest version, I can no longer pin it to my taskbar because of the warning screen thing

That's strange. The warning should only come up one time, the first time you launch Earful. It should remember that you've agreed to it the next time. Do you see the warning every time you start? Also, once you get past the warning, you should be able to pin the icon to the taskbar. Does that not work?

 

[pkane]

I actually tried equal loudless like this a while ago, but couldn't wrap my head around how to apply it to my headphone's FR, given that they themselves are not flat.

Using HD600, starting volume -20:

20, -20
24, -3.5
31, -5.5
39, -8
48, -9
60, -11.5
76, -14
95, -15.5
118, -17.5
148, -20.5
185, -20.5
232, -21.5
289, -21.5
362, -21.5
452, -21.5
565, -22
706, -21.5
883, -20.5
1103, -21
1379, -21
1723, -21.5
2154, -21.5
2691, -22
3363, -22
4203, -22
5252, -22
6563, -22.5
8202, -17.5
10249, -27.5
12808, -21.5
16005, -18.5
19999, -20

Which test were you running here? Two-device EQ or equal loudness curve? (never mind, I see it was equal loudness curve)