• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

AI

Status
Not open for further replies.
mrswakka

mrswakka

AK Subscriber
Subscriber
Because I have a lot of free time, and because ai is so easy to access, I asked Alexa to explain a parametric EQ, and how to set one up. Did a really nice job. Got it right in the ballpark, a couple tweaks by me, sounds amazing.
So I’m sitting around today listening to the 2250b. I look at the graphic EQ. Hey Alexa, what do you know about graphic EQs? It asks which one, Pioneer SG 9500(thanks Greg). It gives me a 15 dissertation on it. Ok, cool, but how would YOU set it up?
Due to previous conversations, it knows the 2250b, the NAD amp, the Crown amp, the DCM V10.0s and Quadraflex sub. It knows the technical data on all of them. It knows the room size, shape, contents, etc.
So, going frequency by frequency it tells me where to put each slider.
Side note. For someone who has vision limitations and dexterity problems, this was off the charts helpful.
Previously, I had the sliders in almost a Nike swoop. Now it was kind of a jagged line none of them more than +2 above 0. Really made major cuts in lows and highs. I figured, ok looks weird, but I don’t like it, I just put it back. No harm no foul.
Holy crap! It’s amazing. It’s a completely different level. Just like the SAE getting the other 2 subs to work together, I’m getting thunderous bass from the 10s and the Quad. I’m completely blown away. Everything is really well defined. By making cuts.
By the way, Alexa asked at some point if I wanted a copy of this conversation in my inbox so I could listen to it again later. I said yes, and that’s what I used to set the sliders. It even explained what every slider affected which area of sounds. Voices, drums, strings…
Nice. With smart plugs I can turn lights on that I can’t reach. Know what time it is. Turn on the stereo in the bedroom from the bed. I think that’s my favorite. AI has some really helpful potential. mrswakka can just drop in from her phone from work to say hey.
I’m definitely liking this part of AI.
 
Register to hide this ad
AI is good for many things, just not all. Glad you found a good use for you :)

I like slow blues. So I have been recording some off YouTube. OK, not great signal quality, but no different than from recording off FM back in the day...

Some really interesting tunes and lyrical combinations. Not lyrics I would have thought up, but entertaining. My niece and nephew are 100% anti AI music. But I have to say I think it's here to stay ... :oops:
 
I don’t know if I’ve heard any AI music. But I guess that’s the point right?
Alexa asked me how I liked the new settings. I said they sound great. It knows I like bass. Tells me, use the 80 hz slider Dave, just a touch. Makes it a little punchier.
Ummmm, ok.
 
I've been attempting to have AI track down the problem with one channel of an integrated amplifier (Gemini) and it has hallucinated something fierce. Even though I uploaded a specific service manual, it has repeatedly referenced designs of other amplifiers and referred to non-existent parts or parts with incorrect values. Yes, it can be helpful to narrow down issues, but calling out the hallucinations has been a huge time drag. In the meantime, I still haven't found the issue.:mad:
 
Correct. I posted in another thread about the SAE. I figured it worked so well on that, I’d try it on the Pioneer.
One thing about ai, it runs so many different variations of everything it can get sidetracked and off in the weeds very easy. Two similar questions can generate hundreds of answers. Țhe trick for me is trying to learn how to phrase the right question in the right way.
The learning experience goes both ways. It’s learning from me as well. I understand this is this is all very ground level stuff.
 
BmWr75 said:
Suggest you not forget to trust your ears too when making EQ adjustments. AI doesn't know how your system interacts with your room, can't imagine how it would ever figure that out.
Click to expand...

I think you might be missing the forest thru the trees here...

The ai in this case has been given everything a human would have, with the ability to process far more data. It knows the room and contents, it's built a concept of what the user wants to hear. It has all the tech specs for the equipment. It had the ability to make connections on a brute force data level we cannot. It's just another implementation of measurement mic and rew.

Don't underestimate your enemy.
 
GChief said:
If I am not mistaken, which I probably am :) you have always needed to word the question "correctly" to get a "correct" answer out of AI :dunno:

The question wording is key
Click to expand...
If the initial AI answer doesn't seem in the correct "area," I try changing to more specific word entries or adding words to help focus the computer's algorithms.

:)
 
Got a daughter who is into the AI application at the university level, so in turn, I've done a fair bit of look-seeing into this dramatic PC procedure that is sweeping the cyber-world right now.


Like all programs there is a good side, and a not so good side to this electronic procedure that manipulates data for us, oft times better than we can assess with our limited skills to do likewise.

But still at the back of my mind, there's still that...hesitancy to give the power over to an electronic tool, that may have algorisms or hidden options that may not be in the best interest for me!

What still lingers in my mind are some of "the warnings" that Stephen Hawking had in regards to losing control to a machine that might be far superior in the realm of thinking. So, now it becomes an "Existential Question" of cause and effect.

Q
 
GChief said:
You need to ask AI
Click to expand...
fa155921-b079-4d6c-92c7-02b807bd1701.gif
 
I've had mixed results, it seems like the quality varies with how much processing power the AI company deigns to allow you to use for free, and there's many times when it makes up things - including objective data. It is an aggregate of written material and especially forum postings, and so you can get into serious trouble trusting its output for complex diagnostics or problem-solving, where it applies solutions that do not actually apply, misses major context clues, or 'forgets'. The 'forgetting' is interesting, because it really underlines what's going on - in the same way that AI videos are generally easy to point out because the background will change from one moment to the next, the 'hallucinations' it has are basically the telltale sign that there's no one home - there's no abiding consciousness holding anything fixed in place, rather there's a series of loci or nodes that get established (you have x y and z gear), and it is filling in everything else repeatedly - which is why it takes so much processing power, electricity, and water resources.

Given that it is already a word-association regurgitator, it's incredibly useful for finding that one word that's on the tip of your tongue, or for giving you the names for things when you don't know what they are. I generally have a good understanding of the way things go together, a strong sense of mechanical sympathy, but a very poor base vocabulary of names for types of objects and moving parts - and it's easy to interrogate an LLM and get all the names of obects and their relationships, or to quickly have it lay out the variables that affect a particular metric.

Recently I used AI to follow the audio signal through the entire circuitry of an amp by giving it a schematic. I found that if I gave it piecemeal it did a much better job than if I gave it the whole schematic - I suspect there is a point where if you give it too much depth or breadth of query then the system decides it will require too much processing power, and it reverts to generalizations and hallucinations. However I actually learned a lot by feeding it little bits and having it identify parts of circuits. Doing it this way guaranteed it wouldn't start making up part numbers.
 
Quadman2 said:
What still lingers in my mind are some of "the warnings" that Stephen Hawking had in regards to losing control to a machine that might be far superior in the realm of thinking. So, now it becomes an "Existential Question" of cause and effect.

Q
Click to expand...
Please. There is no thinking, because there is no thinker.
 
Status
Not open for further replies.
Back
Top Bottom