I found that the free version is adequate for anything I have done so far, now that I am more familiar with entering the necessary data and explaining up front what I want to accomplish. ChatCPT is (based on my very limited experience with it) a great tool because it provides answers and specific suggestions with understandable explanations. I have gotten much information here on the forum, but not in such a condensed and complete format. It may be a good starting point in setting up hearing aids for folks who are new to the DIY process. But, as I say below the Target default values, for me, have been closer to the mark. As a fun project I entered data and asked ChatCPT to provide input values for Target. I entered my audiogram, UCl values and measured feedback thresholds and asked for NHL-NL2 inputs. I got a nice chart and I entered the values. The initial volume recommended by Chat was much too loud. I told Chat about that and It gave me a revised table of input values but the clarity of speech was still pretty bad. For reference, the initial inputs for adaptive Phonak digital created by Target, so far has resulted in more crisp and understandable speech. But the educational value of the explanations along the way have been very valuable.
Yeah this is the problem, itāll be wrong for most people obviously, but the point Iāve been making is the software has already done what AI is giving you, itās already built in, everything you need to make the best effort, so yes I agree that for DIY newbies, AI could help them with the user guides, as in how to use and when to use certain features, but for now I donāt see AI offering anything that the software already has or can be done to get a ābetterā fitting. AI in the very near future is going to dominate a lot of things in our lives for sure, but not quite there yet, a bit scary reallyā¦
It is kind of a fun exercise. When I have time I am going to try Grok and see how close they are to the Chat or Target values. At the end, since I had asked for NAL-NH2, I went back and asked for a chart for Adaptive Phonak 3. Those values were substantially different from Target so I would not recommend that it be used by a new DIY person. Rather I would recommend that they start with the software generated gains (Target in my case) and use the problem solving facilities to fine tune. That being said, if you are willing to go back and forth with the AI software enough times you might finally get a good result.
Iāve done a lot more work on fine tuning the music program with much help from Chatgpt. The latest settings have been devised with lots of context including my audiogram, the infinio Sphere i-90s, my speakers (Sonos Era 300s) and a selection of my favourite tracks. The result has been great in terms of bass and rhythm but vocals arenāt there yet - thin and hissy. Iāve sent an email to my audiologist updating him and with the screenshot. It will be interesting to get his take tomorrow.
Keep at it, will be worth it in the end. Good on you!
I spoke to my audiologist today and he was very positive about how I have used Chatgpt to fine tune the music program on the spheres. In fact, he also uses Chatgpt especially for technical aspects of audiology. He speculates that there will soon be a dedicated audiology AI bot.
That is good to know. It provides reassurance to those who have begun to experiment with using ChatCPT or other programs to optimize fittings.
So I am somewhere near the end of fitting my spheres for listening to music via speakers. Chatgpt has helped a lot. Firstly in helping me find my way around Target. Secondly in suggesting gain levels and frequencies to tweak based on my feedback of how different tracks sound to me. I did all the heavy lifting but I wouldnāt have got far without Chatgpt. My modus operandi is to spend a couple of hours listening to favourite tracks and noting down issues, eg need warmer bass. Consult Chatgpt for one key issue. Tweak the relevant gain levels and live test before saving with 3 test tracks. Save with a note about what I tweaked and with what purpose. Rinse and repeat.