(This Post was inspired by https://www.audiosciencereview.com/forum/index.php?threads/hearing-loss-a-hobby-fades-away.7316/)
A couple of weeks ago I embarked on a project that I like to share and ask for comments: Ear or Hearing EQ.
After having toyed with Room EQ, I thought it would be cool to have the DSP-EQ correct for my hearing deficiencies too: I am close to 60 and have a tinnitus.
And while I could of course get my last medical´s audiogram, I thought it would be better to measure my hearing in my room with my equipment and hence merge Room and Ear EQ.
The goal is the definition of a new room-curve, merging Toole´s with my hearing-deficiency.
My setup is a NanoAVR HDA-DL with DIRAC, an Umik-1 and a PC running the Dirac GUI and REW. The NanoAVR´s analog output feeds into an analog pre-amp that feeds 3*2ch amps. I have a 5.1 setup with all identical speakers but use only left and right at this time for measurements.
I have the UMIK pointing down on me, 20 cm above my head (using 90deg calibration) and run the REW Signal generator and the RTA display. I adjust the volume of my preamp to check my hearing threshold using discrete sine-waves from 20-18000hz in induvial steps. I note down the RTA DB Level when I start hearing the tone. Dirac is on but the filters are off, meaning for NANOAVR that the levels are still matched. I do this for the left and right ear separate and averaged, I move my head to catch high frequency beams. I invite my family and friends for this too. I average 3-5 measurments runs for plotting.
While I initially believed that my tinnitus is the problem, I discovered that there is more. At first I used the ISO 226-1987&2003 threshold data for comparison to make sure I do not measure nonsense. My low frequency hearing matched the ISO curves quite well. Above 10Khz it was different and probably reflects loudspeaker vs headphones and my age. And then I was surprised seeing my hearing response starting at 2 Khz stretching with a loss of 10 dB to 6 Khz. I measured young people and well-aged ones for comparison.
.
I searched the Internet and found literature that shows, that my hearing response seems to be not un-usual for my age group.
I could match my younger peer groups hearing to published 20-30 year´s hearing which gave me confidence that I am not measuring nonsense. I used that “20-30year” curve as a baseline where I like to correct my hearing to, and created a “difference” curve. I gave up measuring in the 2000-200 hz range as my room was too noisy (20-30dB) to really record the threshold there. Above 2 Khz and below 200 hz measurements seemed valid.
But I only measure the hearing threshold, not the effective hearing at 60-80 dB where I usually listen to music. I did not want to transfer my “difference” curve from the 20-25 dB threshold to my 80dB music world. I regressed a transfer function based on ISO226-2003 to shift from 20 to 80 dB and modified my "20 dB-difference" curve accordingly to a "80 dB-difference" curve. I looked into the problem, that those ISO curves were measured with young people and that these loudness responses change with age: there is one publication on that and this is for future enhancements…
I then subtracted the Toole room curve from my “difference” curve and used the results as new target curve in DIRAC (realizing that a room curve is probably “db”-sensitive too…). DIRAC cannot do magic and could be limited by the NanoAvr´s DSP implementation too. Checking with REW, I realized that I could not get Dirac to do any correction I wanted. The finesse in the 5-8Khz range could not be reproduced. So I overdid the complexity in the design not realising that the actual world is different. I went forth and back between Dirac and Rew to check the possibilities. DIRAC also checks what the speaker seems to be capable of and does not design a filter that seems to fall outside of the Speakers-comfort zone. And after I had modified my NanoAvr to Dirac, I sadly cannot simply go back and try some REW filters.
For the low frequency I am still not settled. I noticed that the perceived loudness difference between 100-60db quite closely matches Toole´s observation for untrained people – that sounded like a rational basis to do it and I use it but this is a bit bass-heavy. To be improved.
I then tried to create separate target curves for my left and right ear but am still working on that. My right speaker´s tweeter did not like my experiments with 16Khz at 85 dB, so there is now also a tinnitus build into my right speaker and I have to fix that first before fine tuning left and right.
The results – are great. Well, one could say lifting the mid-heights and the bass (as my workflow´s byproduct) always sounds nice and pleasant. The question could be asked if what I perceive as ”hearing as I where young” is simply an illusion. Sure. But nevertheless, I think it is fantastic, I do enjoy music played like that (my young peer groups thinks it is too bright) and believe that this sounds as I heard music many years ago. Again, could be an illusion but then it is a nice one.
I do believe that this method has a future: After doing room EQ, why not checking on the hearing of the user too and allow an option for “personal optimization”?Looking at the published data, many people above 45 years old suffer from a frequency response loss above 3Khz, the loss above 10khz is rather trival. So many people would benefit. And there might be even a market for a strong EQ-DSP and liquid cooled tweeters that could handle that..
The next goals is the head transfer function- my surround speakers are only 10-20deg behind my head, so they “shine” right onto my ear-drums. I notice that my hearing correction is too bright there as it was optimized for front left and right. And there is the center too.
Hoping for good comments.....
References:
F Toole: The Measurement and Calibration of Sound Reproducing Systems. J Audio Eng Soc, Vol 63, No 7/8, 2015
AR Valiente, AR Fidalgo, JR Garcia-Berrocal, RR Camacho: Hearing threshold for an ontologically screened population in Spain. International Journal of audiology, April 2015
K Kurakata, T Mizunami, K Matsushita: How much is the individual difference in hearing sensitivity. Acoust. Sci & Tech 34, 1 (2013)
H Moller, MF Sorensen, D Hammershoi, CB Jensen: Head-related transfer functions of human subjects. Journal of the Audio Engineering Society, 43(5), 1995.
https://www.nonoise.org/library/handbook/
A couple of weeks ago I embarked on a project that I like to share and ask for comments: Ear or Hearing EQ.
After having toyed with Room EQ, I thought it would be cool to have the DSP-EQ correct for my hearing deficiencies too: I am close to 60 and have a tinnitus.
And while I could of course get my last medical´s audiogram, I thought it would be better to measure my hearing in my room with my equipment and hence merge Room and Ear EQ.
The goal is the definition of a new room-curve, merging Toole´s with my hearing-deficiency.
My setup is a NanoAVR HDA-DL with DIRAC, an Umik-1 and a PC running the Dirac GUI and REW. The NanoAVR´s analog output feeds into an analog pre-amp that feeds 3*2ch amps. I have a 5.1 setup with all identical speakers but use only left and right at this time for measurements.
I have the UMIK pointing down on me, 20 cm above my head (using 90deg calibration) and run the REW Signal generator and the RTA display. I adjust the volume of my preamp to check my hearing threshold using discrete sine-waves from 20-18000hz in induvial steps. I note down the RTA DB Level when I start hearing the tone. Dirac is on but the filters are off, meaning for NANOAVR that the levels are still matched. I do this for the left and right ear separate and averaged, I move my head to catch high frequency beams. I invite my family and friends for this too. I average 3-5 measurments runs for plotting.
While I initially believed that my tinnitus is the problem, I discovered that there is more. At first I used the ISO 226-1987&2003 threshold data for comparison to make sure I do not measure nonsense. My low frequency hearing matched the ISO curves quite well. Above 10Khz it was different and probably reflects loudspeaker vs headphones and my age. And then I was surprised seeing my hearing response starting at 2 Khz stretching with a loss of 10 dB to 6 Khz. I measured young people and well-aged ones for comparison.
.
I searched the Internet and found literature that shows, that my hearing response seems to be not un-usual for my age group.
I could match my younger peer groups hearing to published 20-30 year´s hearing which gave me confidence that I am not measuring nonsense. I used that “20-30year” curve as a baseline where I like to correct my hearing to, and created a “difference” curve. I gave up measuring in the 2000-200 hz range as my room was too noisy (20-30dB) to really record the threshold there. Above 2 Khz and below 200 hz measurements seemed valid.
But I only measure the hearing threshold, not the effective hearing at 60-80 dB where I usually listen to music. I did not want to transfer my “difference” curve from the 20-25 dB threshold to my 80dB music world. I regressed a transfer function based on ISO226-2003 to shift from 20 to 80 dB and modified my "20 dB-difference" curve accordingly to a "80 dB-difference" curve. I looked into the problem, that those ISO curves were measured with young people and that these loudness responses change with age: there is one publication on that and this is for future enhancements…
I then subtracted the Toole room curve from my “difference” curve and used the results as new target curve in DIRAC (realizing that a room curve is probably “db”-sensitive too…). DIRAC cannot do magic and could be limited by the NanoAvr´s DSP implementation too. Checking with REW, I realized that I could not get Dirac to do any correction I wanted. The finesse in the 5-8Khz range could not be reproduced. So I overdid the complexity in the design not realising that the actual world is different. I went forth and back between Dirac and Rew to check the possibilities. DIRAC also checks what the speaker seems to be capable of and does not design a filter that seems to fall outside of the Speakers-comfort zone. And after I had modified my NanoAvr to Dirac, I sadly cannot simply go back and try some REW filters.
For the low frequency I am still not settled. I noticed that the perceived loudness difference between 100-60db quite closely matches Toole´s observation for untrained people – that sounded like a rational basis to do it and I use it but this is a bit bass-heavy. To be improved.
I then tried to create separate target curves for my left and right ear but am still working on that. My right speaker´s tweeter did not like my experiments with 16Khz at 85 dB, so there is now also a tinnitus build into my right speaker and I have to fix that first before fine tuning left and right.
The results – are great. Well, one could say lifting the mid-heights and the bass (as my workflow´s byproduct) always sounds nice and pleasant. The question could be asked if what I perceive as ”hearing as I where young” is simply an illusion. Sure. But nevertheless, I think it is fantastic, I do enjoy music played like that (my young peer groups thinks it is too bright) and believe that this sounds as I heard music many years ago. Again, could be an illusion but then it is a nice one.
I do believe that this method has a future: After doing room EQ, why not checking on the hearing of the user too and allow an option for “personal optimization”?Looking at the published data, many people above 45 years old suffer from a frequency response loss above 3Khz, the loss above 10khz is rather trival. So many people would benefit. And there might be even a market for a strong EQ-DSP and liquid cooled tweeters that could handle that..
The next goals is the head transfer function- my surround speakers are only 10-20deg behind my head, so they “shine” right onto my ear-drums. I notice that my hearing correction is too bright there as it was optimized for front left and right. And there is the center too.
Hoping for good comments.....
References:
F Toole: The Measurement and Calibration of Sound Reproducing Systems. J Audio Eng Soc, Vol 63, No 7/8, 2015
AR Valiente, AR Fidalgo, JR Garcia-Berrocal, RR Camacho: Hearing threshold for an ontologically screened population in Spain. International Journal of audiology, April 2015
K Kurakata, T Mizunami, K Matsushita: How much is the individual difference in hearing sensitivity. Acoust. Sci & Tech 34, 1 (2013)
H Moller, MF Sorensen, D Hammershoi, CB Jensen: Head-related transfer functions of human subjects. Journal of the Audio Engineering Society, 43(5), 1995.
https://www.nonoise.org/library/handbook/