oluvsgadgets
patreon
Why does tuning take so long?
🕑 Added 2022-02-28 11:55:10 +0000 UTCI want to apologise, but I won't manage to finalise the current Free Pro 2 tuning with all resonance frequenices for February. I was fighting to optimise the settings since RC2 but it will take even more effort to finish everything. Meanwhile I got several messages asking if there was a way to revert to the default factory tuning. Therefore I prepared a very quick and rough setting, which should simulate the original response of the Free Pro 2 more or less. This setting is purely based on measurements, I haven't even listened to the result, but the measurements are close enough I would say. Black is the original tuning measured with a new unit which I just opened, while blue is the simulated tuning measured with my reference unit:
This took me about half an hour to prepare and was not such an effort:
On the other hand creating a tuning which should really sound "correct" without added colorations is way more effort, therefore let me explain why:
My reference is pink noise listened over various sources. I am trying to recreate the perceived spectrum as a whole, which already is not that simple just by listening. The human ear will quickly adopt to small errors which makes frequency matching a trial&error procedure. On top of this due to its complex spectrum pink noise obviously tends to create artefacts when listened over my PC, probably a result of the Bluetooth codec. Sometimes it's worse, other times better, but it's hard to judge the result properly, therefore I need to transfer my latest settings into the Airoha app on my Android phone where pink noise will suddenly again sound different. My Android phone is quite ancient and it causes repeated Bluetooth cut outs, which again changes the timbre every time this happens, therefore I still cannot judge the sound correctly and need to finally listen over my iPhone to really hear the correct result of my latest settings. If I am somehow satisfied I need to let this setting settle and usually listen to it for a day or two until I may notice some further issues and need to get back to my PC again and work from there readjusting the settings accordingly still hopefully remembering what issue I was hearing on my iPhone which played back over the PC will again sound different. It's quite an exhausting procedure as I try to finetune for the smallest detail to have the most correct result possible. Sometimes I need to add additional filters to correct for the flaws I am hearing. Then I measure the result and try to recreate that response with only the 10 available filters. As all filters interfere with each other, adjusting one filter, will affect the others too, this makes the entire procedure even more complicated. Only if I personally manage a result which I think is valid, I can measure the final response to get a reference curve which hopefully is correct and which I can use as base for creating all further tunings for the other resonance frequencies. I usually tune with a "reference unit" which has a quite consistent response for left and right, but I also try out my latest settings on my other units to notice that there are again some issues and I will try to find values which will offer a better result among the other units as well still hoping that it will sound also OK on my reference unit. It's a painfully frustrating experience since I started all my tuning work, and I am starting to question everything, but at the same time I have the impression as if the results were becoming better and better and I think it's definitely worth the effort.
I made some progress since RC2 and the resulting graph became even more "simple" than it was before, suggesting that I am on the right path. The 7k dip became a part of the final response and is not a real dip anymore. The correct ratio between different frequencies seems to be the secret, but I still haven't figured out how exactly I need to proceed to get the desired effect as I am still reyling mostly on my ear and only measure if I think the current setting may be "correct" to some degree.
I wanted to show you a measurement of the current status in black compared to RC2 in blue and my original neutral tuning in green:
They do not differ that much from the graph, but actually the correct ratio between upper mids and treble is crucial to make the result sound "correct". All of these settings do not sound bad, but I try to really achieve a neutral and uncolored representation which will help to offer you also a valid sound for my upcoming binaural recordings. "Neutral" may not be to everyone's taste, but I think it's the most important tuning and a starting point for any further sound modelling approaches.
When sweeping throughout the frequency range I get a more constant amplitude now with only small variations at 6,7,9khz. I am still trying to get this as smooth and correct as possible, and without continuous comparisons to my references the result may drift into some wrong direction. I have another several dozens of variants in the meantime since I started "newtral".
Another big issue is that the measuring couplers were never designed to measure an average human ear canal correctly. They were designed with hearing aids in mind which are usually inserted way deeper, thus resonances like 8-13khz are achieved. 13Khz is regarded as "standard" for a IEC711 coupler, but IEMs with a more shallow insertion create resonances at way lower frequencies, which the couplers obviouisly do not measure properly. Not only the measured bass amount is wrong but also the amount of mids will change depending on the insertion depth and I have no idea how to deal with this.
Below is a measurement of my latest setting shown with a 6khz resonance in black and inserted deeper to 7khz in blue just adjusting the 6K filter to 7K leaving everything else the same. You can see that the entire response changed, not only bass is lower now and is probably measured more correctly with the 7K resonance, but mids became lower in amplitude as well. With 6K I am above the Harman target, while with 7K and the same settings I am below the target. I have no idea if the 7K measurement shows this range more correctly or if it needs to be adjusted to the same level as the 6K measurement to be perceived the same sounding.
All this is very vague and shady, and if I fool around for days to find a setting which I regard as "neutral" which hardly differs from my previous settings in a measurement, but then look at all the weird responses from IEMs that cost several 1000$ which all look completely different, as if they were painted by random, I simply cannot understand this entire craziness:
I will need some further days to tweak my final "newtral" settings, which hopefully will work well among other units as well. I am also planning to rebuild my filter values as the current setting results in quite a low output volume, but I will also need your help, preferrably from some trained ears with higher earcanal resonances than mine, who can help sorting out the measuring issues I pointed out above, only then I can start my work on all the other settings for different ear canal resonances, but then we will hopefully achieve a "reference" tuning which should also satisfy more demanding listeners.
I wanted to thank you all for your continuous support, although I am slowly getting tired of all this, I am still confident to achieve something particular.
