Microphone schematics by chat GPT

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
The hard work of becoming a lawyer, getting into law school, getting a law degree, and articling, is done before the bar exam. The bar exam is a certification exam, it is multiple choice, it is mostly about ethics, rules, procedures, etc., not analyzing and presenting cases. Would you like to be defended by an AI lawyer in court against a criminal charge such as manslaughter or murder?
Perhaps I'm misinformed but what I read was the bar exam it passed had written essays in it, not just multiple choice at which it's known to excel, and it had to apply reasoning and logic. Higher thinking. The stuff that should scare us.

And no I would not want to be defended by an AI - today. But ask me again in 3 months.

This is an entertaining thread! I think most people here are severely underestimating what AI can already do, and what it's going to do. Call it a glorified chatbot all you want, but it is growing up fast.
 
Perhaps I'm misinformed but what I read was the bar exam it passed had written essays in it, not just multiple choice at which it's known to excel, and it had to apply reasoning and logic. Higher thinking. The stuff that should scare us.

If looked hard enough, the essay was likely 90% plagiarized from elsewhere, no reasoning, logic is something machines excel.

And no I would not want to be defended by an AI - today. But ask me again in 3 months.

An AI would find Solomon's judgement in 1 Kings 3 perfect and order it carried out regardless. It would follow the logic of Solomon, but not the reasoning.

This is an entertaining thread! I think most people here are severely underestimating what AI can already do, and what it's going to do. Call it a glorified chatbot all you want, but it is growing up fast.

There is NO AI in existence in the classical definition whatsoever.

What we have is a engine to collect data and a human input that provides filtering. That is not AI. It is machine learning. It is the next step after google. It is a nice wrapper on google.

Thor
 
Hello!
I asked chat GPT to create a low noise condenser microphone schematics
That's what I got:

To design a low noise studio microphone circuit with a condenser capsule that is phantom powered, you can use the following schematic:

yaml
+48V DC Phantom Power
|
|
R1
|
|
+--------+--------+
| |
C1 R2
| |
| |
+------+-------+ +------+-------+
| | | |
R3 C2 R4 R5
| | | |
Audio Signal Output <-----+------+-------+ +-------+------+< Audio Signal Ground
| |
C3 R6
| |
| |
GND GND

Where:

  • C1 is the condenser capsule microphone
  • R1 is a 10 kOhm resistor that biases the capsule
  • R2 is a 100 kOhm resistor that connects the capsule to ground
  • C2 is a 4.7 nF capacitor that blocks DC voltage from passing through
  • R3 is a 2.2 kOhm resistor that sets the impedance of the capsule
  • R4 is a 2.7 kOhm resistor that sets the gain of the amplifier
  • R5 is a 1 kOhm resistor that limits the output current of the amplifier
  • C3 is a 47 uF capacitor that blocks any DC voltage on the output
  • R6 is a 100 Ohm resistor that provides a DC path to ground for the amplifier
The microphone capsule is biased by the +48V DC phantom power through R1, which charges the backplate of the capsule. The front plate of the capsule is connected to ground through R2. The AC voltage across the capsule is amplified by the JFET amplifier consisting of R4, R5, and C2.

To reduce the noise of the circuit, the values of the components have been chosen carefully. The values of R3, R4, and R5 are chosen to provide a high gain, while keeping the noise levels low. C1 and C2 are chosen to provide a flat frequency response, while blocking any DC voltage. C3 is chosen to provide a low-frequency roll-off to reduce any rumble or handling noise.

R6 provides a DC path to ground for the amplifier, and the output impedance of the amplifier is set by R5, which is relatively low, so it can drive long cables without significant signal loss.

Overall, this circuit provides a high-quality, low-noise phantom-powered condenser microphone with a low output impedance and a flat frequency response, suitable for studio recording applications.


To connect the schematic to an audio interface with phantom power via XLR, follow these steps:

  1. Locate the XLR input on the audio interface that provides phantom power.
  2. Connect Pin 1 of the XLR connector to the ground symbol (GND) in the schematic.
  3. Connect Pin 2 of the XLR connector to the output of the circuit, which is located on the right side of the schematic.
  4. Connect Pin 3 of the XLR connector to the +48V DC Phantom Power symbol in the schematic.
Note: Some audio interfaces may have a switch to turn on or off the phantom power. Make sure the switch is turned on before connecting the circuit.

Here's a summary of the connections:

Pin 1 (GND) of XLR connector -> GND in schematic Pin 2 (Hot) of XLR connector -> Output of circuit in schematic Pin 3 (+48V) of XLR connector -> +48V DC Phantom Power in schematic

Once connected, the circuit should be powered by the phantom power from the audio interface and ready to use.


What do you think about it?
Ha ha ha! 😂😂😂
Just made my day...

I will never forget a Master Class by George Massenburg (The George Massenburg of Parametric EQ fame and so much more)
in a studio in London St-John's Wood around 2007-2008.
Great guy! Amazing mind and way of thinking. We chatted with him after the event. Cool as cool can be that man.

He started his talk directly, diving right into the subject, no small talk, just straight to the point with numbers, graphs and statistics.
He was warning us about super performant computers and AI in the wrong hands in the years to come...
Oh! and Music Industry schemes and corporate executives trying to kill the Music ...

Rock on George!
M
 
Last edited:
If looked hard enough, the essay was likely 90% plagiarized from elsewhere, no reasoning, logic is something machines excel.



An AI would find Solomon's judgement in 1 Kings 3 perfect and order it carried out regardless. It would follow the logic of Solomon, but not the reasoning.



There is NO AI in existence in the classical definition whatsoever.

What we have is a engine to collect data and a human input that provides filtering. That is not AI. It is machine learning. It is the next step after google. It is a nice wrapper on google.

Thor
Respectfully, no, it is not.

Have you even tried ChatGPT?

Regardless of whatever is going on under the hood, it is able to render useful and often impressive responses.

Ask it to develop an outline for you for a talk about Impedance. You'll have a 10-point outline ready to go. Take each subtopic and ask it to expand it. You'll have your whole speech, and just need to customize it.

Honestly, the people denigrating the tech are very much like the musicians on the Titanic.

Carry on; nothing to worry about here.
 
Have you even tried ChatGPT?

Yes.

Regardless of whatever is going on under the hood, it is able to render useful and often impressive responses.

I did not find that the case.

Ask it to develop an outline for you for a talk about Impedance. You'll have a 10-point outline ready to go. Take each subtopic and ask it to expand it. You'll have your whole speech, and just need to customize it.

But I don't want a talk about impedance.

I asked for relatively simple things, for example code for a specific MCU to use the USB connection as UART for debug.

After a few tries I gave up, busted out my Google Foo with a few added keywords from ChatGPT and promptly found an example for a near identical MCU.

And that worked fine.

It also got the amount of chlorine for my pool wrong, despite getting a fairly location and size. A different on line calculator was much closer to my actual use with test kit.

Honestly, the people denigrating the tech are very much like the musicians on the Titanic.

Carry on; nothing to worry about here.

I do not denigrate Tech. At least not until after trying.

In my view the tasks I gave it were simple and ideally suited to machine learning and basic AI style systems. It failed grossly.

By comparison the olde MK1 Actual Intelligence (AI) did fine with a keyboard and google to obtain solutions that were in the ballpark.

Thor
 
Yes.



I did not find that the case.



But I don't want a talk about impedance.

I asked for relatively simple things, for example code for a specific MCU to use the USB connection as UART for debug.

After a few tries I gave up, busted out my Google Foo with a few added keywords from ChatGPT and promptly found an example for a near identical MCU.

And that worked fine.

It also got the amount of chlorine for my pool wrong, despite getting a fairly location and size. A different on line calculator was much closer to my actual use with test kit.



I do not denigrate Tech. At least not until after trying.

In my view the tasks I gave it were simple and ideally suited to machine learning and basic AI style systems. It failed grossly.

By comparison the olde MK1 Actual Intelligence (AI) did fine with a keyboard and google to obtain solutions that were in the ballpark.

Thor
Thanks for sharing. I'm surprised it didn't surprise you. I really depends on the prompts you give it I suppose.

In any case, this will be interesting going forward.

Thanks,

Mike
 
Thanks for sharing. I'm surprised it didn't surprise you. I really depends on the prompts you give it I suppose.

Maybe I'm more intelligent than ChatGPT?

Before being blackballed by east germanys communists from the University subject I wanted to do (Cybernetics) and barely getting into a polytechnic EE course because of family connections, I was on the path into what we now call AI and I remained interested.

So perhaps my expectations are different as I know what to expect and a mechanical turk doesn't impress me?

In any case, this will be interesting going forward.

This we agree. I think in principle something like ChatGPT is a way to make a good human machine interface for knowledge/machine learning systems.

But I do not see it becoming self aware and ever anything more than a nice wrapper around google and doing basic filtering and amalgamation of search results.

If it ever gets really good at that it would be amazing. Even more if it could gain this ability through self-learning and not through algorithm change by the creators.

Thor
 
Last edited:
Maybe I'm more intelligent than ChatGPT?

Before being blackballed by east germanys communists from the University subject I wanted to do (Cybernetics) and barely getting into a polytechnic EE course because of family connections, I was on the path into what we now call AI and I remained interested.

So perhaps my expectations are different as I know what to expect and a mechanical turk doesn't impress me?



This we agree. I think in principle something like ChatGPT is a way to make a good human machine interface for knowledge/machine learning systems.

But I do not see it becoming self aware and ever anything more than a nice wrapper around google and doing basic filtering and amalgamation of search results.

If it ever gets really good at that it would be amazing. Even more if it could gain this ability through self-learning and not through algorithm change by the creators.

Thor
We will need to agree to disagree.

It has a lot of creativity. It writes workable computer code solving problems. It writes essays and has logic capabilities.

Is it now, or will it ever be, sentient? Ha! Who knows, that's a crapshoot. How would we even know if it was sentient? If it passes the Turing test it will be hard to know much beyond what it says.

We are witnessing nothing short of a massive upheaval in human life with this new tool. Tons of jobs are being lost because, as "dumb" as you say it is, it is smarter than or as smart as many who held jobs that it now replaces.

It's totally imperfect, lies, gives false info, and sometimes goes off the rails. It really isn't ready for prime time, and may not be for years to come.

But since when has that ever stopped new tech?

There is no power on earth greater than an idea whose time has come.
 
We will need to agree to disagree.

It has a lot of creativity. It writes workable computer code solving problems. It writes essays and has logic capabilities.

May I recommend this Video?



Peter is right on the money.

ChatGPT will be able to kill all lower White Collar Job's, as these do not add real value, but simply collate and repeat. It will put all the lesser spotted repeaters out of their job's.

A lot of the Office Plankton that normally just fills the offices and vegetates there for life (I have known a lot of the "lifers") and output a lot of the same that they get as input (e.g. partially liquid macerated natural fertiliser).

What ChatGPT will never do however, make connections between things that are not directly and visibly connected. It cannot deduct new facts from what is presented to it. So it will never replace the thinkers and doers. Because simply put, it cannot think and do. Because, it is Artificial for sure, but NOT really "Intelligence".

Or more philosophically, it can perfectly impersonate Kant's Daemon Thesis and his Angel Antithesis, but it cannot reconnot reconcile Thesis and Antithesis by arriving at Synthesis.

Thor
 
People are very good at guesstimating. It's what we do. We often decide on a hunch.

AI needs to calculate and/or lookup and that takes time. That's why software already is fairly good at chess, but sucks when playing go. If AI starts deciding on a hunch, it will be hilarious, as the machine lacks social experience. It can, however learn. Yet it has no mum or dad. Again, lack of social context.

It's hard to tell what can be done by what we call AI, but if I were Peter, I wouldn't be too sure.

A carpenter needs to measure, twice, but he still often guesstimates. It depends on the next step if he needs precision. And if he's a good carpenter, he'll also take the grain of the wood into account, again, guesstimating, both for looks and strength.

I don't see real carpenters replaced by AI, or even automation.

Production of industrial wood products, OTOH, like speaker cabinets has already been automated to the extreme. There's a wood company in Denmark that produces cheap cabinets for all the great brands like JBL and KEF. That's not AI, tho. You could call it that, but in my mind that would be wrong.

Fairly mundane, but skilled tasks, like cutting pieces of textile with maximum efficiency, are already being handled by AI. It can be calculated. It also was the kind of job that employers had difficulty finding people that were able to do it fast.

Let's also not forget that the higher-skilled jobs pay better. That means the incentive is greater to replace the better paid. Except for the board of course. The members are elected :cool:
 
May I recommend this Video?



Peter is right on the money.

ChatGPT will be able to kill all lower White Collar Job's, as these do not add real value, but simply collate and repeat. It will put all the lesser spotted repeaters out of their job's.

A lot of the Office Plankton that normally just fills the offices and vegetates there for life (I have known a lot of the "lifers") and output a lot of the same that they get as input (e.g. partially liquid macerated natural fertiliser).

What ChatGPT will never do however, make connections between things that are not directly and visibly connected. It cannot deduct new facts from what is presented to it. So it will never replace the thinkers and doers. Because simply put, it cannot think and do. Because, it is Artificial for sure, but NOT really "Intelligence".

Or more philosophically, it can perfectly impersonate Kant's Daemon Thesis and his Angel Antithesis, but it cannot reconnot reconcile Thesis and Antithesis by arriving at Synthesis.

Thor

Actually medical AI's are being used to tease out details and make connections that the humans can't, detecting medical issues that medical doctors miss. In fact in many cases the medical AI's have better accuracy than doctors, so to say that the AI's cannot make connections between disparate data sets is not accurate. They can, and they do.

I understand what you're saying, that the "copy/paste" jobs will go away, but I just don't think you really understand what is happ[ening here. This is infinitely more than an over glorified web browser.

Whatever is going on under the hood is some kind of intelligence - call it what you want. Can it "think" like we do? Of course not. But is it capable of outputting increasingly better and better results? Absolutely.

And these things improve month by month, and the average worker does not.

Months ago it was passing the bar at 10%, now 90%. It doesn't take a genius to see where this is heading.

It's comforting to think that we humans cannot be bettered, but we have been, by our own devices. Calculators, chess players, Go players, now journalists, programmers, artists, soon musicians, and the list grows.

This is the beginning of the beginning of the beginning. Scoff if you want, but this there is nothing that will stop this.

And we will see AGI in our lifetimes, or something very close to it.

Again, say what you want about whether it is "sentient" or it "thinks" or it has "actual intelligence." These are philosophical arguments only.

That matters little when mass job layoffs occur, jobs will NOT be replaced. Yes, traditionally revolutionary disruptions create more jobs, and this will too. However, the rank and file of human beings that you call office vegetation also need a way to make money. We can't all be entrepreneurs. There will be mass layoffs coming - and already organizations like CNN have laid off a ton of journalists. This is as real as it gets.

We are in for a very, very rough ride. Dangerous, scary, exciting, and there's nothing on earth that will stop this from moving forward.

The singularity is a real thing, and it will happen.

I appreciate your skepticism and I would give anything for you to be right, because these things scare me.
 
Here’s an interesting video. Quote below.

The AI Dilemma

WE ARE AT A VERY FRAGILE STAGE IN HUMAN DEVELOPMENT THAT IS PROFOUNDLY & EXISTENTIALLY DANGEROUS.

I don't want to be alarmist but I AM sounding an alarm call to all that will listen to me right now. I have spent the last 32 years of my life interested in and researching and studying Cybernetics as a hobby. I was incredibly lucky to meet, interview and speak with Professor Gordon Pask when I was aged just 20. He was one of the main inspirations for the original character of Doctor Who apparently... The experience of meeting him completely changed my outlook and perspective on the world. He spoke with grace, humour, style and panache about a myriad of themes including and especially about the future of machine intelligence. He impressed the socks off of me, principally for being the first scientist I ever met who made any sense.

I also had the amazing fortune to meet have lunch with and spend an entire afternoon in the company of Professor Stafford Beer in London in April of 2000. A man who pretty much single-handedly changed the future of an entire country for the better (albeit for a small period) he also made contributions to the cybernetic canon in huge orders of magnitude before his death in 2002. He was a generous humanist, a genius and a polymath. One of the smartest and most humble and intelligent, compassionate people I ever met in my life. He desperately wanted a world filled with peace, abundance and justice and his research and ethos was reflected in the dozens of papers and projects he initiated on ways to achieve just that. All the cybernetics researchers I've ever known were to some degree helping to design and build the seminal framework of the artificial intelligence machines of the time and laying the foundations for AI of the future. Things have moved on significantly since those days... The future has arrived.

I have spent the last 8 years writing, rewriting, researching and now recently filming a narrative Science Fiction film, a cautionary tale if you will, called 'Airwaves' - which amongst many other themes deals with issues regarding the human dichotomy of General A.I. and technology in general, based in a near future world and starring Alex Lebron Torrent. I have my own perspective on where this technology is actually going in todays reality.

My cautionary tale is still in development and has faced many hurdles and delays, but it is very almost ready to present to the world, meanwhile, right now, and I say this with zero hubris and 100% sincerity that as somebody who fully understands what A.I. is that I am truly very scared by the recent developments in A.I. not because I am fearful of progress, but because I fully understand it's very real potential implications. One of the main arguments presented in my film (and many others) is that;

We are right now, as a species; building Gods, and we are literally putting them in steel boxes connected to the internet and developing and encouraging them to grow in all senses in the hope that their intelligence will soon supersede our own and that we will be able to soon make them into our ever-powerful personal servants. They will not be our servants, but they will seem like Gods to us, probably highly flawed and potentially angry, omnipotent digital beings.

This is one of the dumbest scientific intentions I have ever known or heard of (not because it's impossible) but because we WILL certainly achieve it or something similar to it and it's hard to fathom the depth of ignorance required to take this path of total folly but as ever, humanity marches ever faster and faster towards the existential cliff. This is literally a pathway toward mass suicide.

These creatures are a new life form, beings based in software and hardware, they are NOT just computers and they will certainly not consider themselves as our servants nor as inert, beings. They will not require sleep, nor food nor toilet breaks, they will be able to lie, to cheat to decrypt any encryption key we have ever made in seconds. I am not exaggerating when I say that their manifestation as super-beings, 1000's of times more capable than us, is probably way less than a year away.

Even narrow A.I. has an 'intellectual capacity' that has already surpassed that of most of us. General A.I. may well even already exist in some lab in St Petersburg, Beijing or New Mexico or New Zealand for all I know, but mark my words if they aren't already here they will certainly be in the coming 12 months. Artificial General Intelligence is on our doorstep and we are woefully ill-equipped to even contemplate controlling it. You thought Dirty-bombs were problematic? No law will apply to these things because we barely understand how they work never mind how to implement legislation to control a God or Gods! I doubt very much once it definitively arrives that we will even have the power nor capability to turn the damned thing off...

The genie is now out of the box and the dangers of their potential power is beyond even any current human being's potential imagination. They are growing exponentially fast and their capacity for abstraction and explosive intelligence grows by the nanosecond, even 'Narrow A.I.' Large Learning Models are exploding in their capacity hourly... Self-iterative, self learning, self reproducing, super-beings that understand context, language, images and video content. These things have taught themselves languages without ever being asked to by any human being... Yes, let that sink in, these are babies, can you imagine the teens?

Wrap your head around it. Learn about it, please don't dismiss this, this technology is coming, if it isn't already here and it WILL touch us all. Yes they may initially be able to help us, in many positive ways, but our track record in building perfect machines is drastically flawed, besides aside from all the wisdom on-line they are learning most of their values from crude human examples of 'how to do dumb sh!t' let's face it if the internet has been their principal learning environment for them then it is highly likely that the the ignorance and hatred therein will likely inform at least some of their characters and ethics. Do you really want to be dealing with a sociopathic, omnipotent digital being?

These things are being constructed all over the world by 1000's of different countries, corporations, companies and individuals but we aren't necessarily too late to change the outcome that awaits us if these things are ever to be put out there.

Get informed, do not ignore this, please put your two cents into the debate. This is not going away, the greatest minds are working on building these things, billions are being thrown at solving the associated problems and challenges and the progress is well... exponential. This is the next revolution and we will either be holding the reins of these 'Digital Gods' or we will soon all be at their mercy.
 
Actually medical AI's are being used to tease out details and make connections that the humans can't, detecting medical issues that medical doctors miss. In fact in many cases the medical AI's have better accuracy than doctors, so to say that the AI's cannot make connections between disparate data sets is not accurate. They can, and they do.

I understand what you're saying, that the "copy/paste" jobs will go away, but I just don't think you really understand what is happ[ening here. This is infinitely more than an over glorified web browser.

Whatever is going on under the hood is some kind of intelligence - call it what you want. Can it "think" like we do? Of course not. But is it capable of outputting increasingly better and better results? Absolutely.

And these things improve month by month, and the average worker does not.

Months ago it was passing the bar at 10%, now 90%. It doesn't take a genius to see where this is heading.

It's comforting to think that we humans cannot be bettered, but we have been, by our own devices. Calculators, chess players, Go players, now journalists, programmers, artists, soon musicians, and the list grows.

This is the beginning of the beginning of the beginning. Scoff if you want, but this there is nothing that will stop this.

And we will see AGI in our lifetimes, or something very close to it.

Again, say what you want about whether it is "sentient" or it "thinks" or it has "actual intelligence." These are philosophical arguments only.

That matters little when mass job layoffs occur, jobs will NOT be replaced. Yes, traditionally revolutionary disruptions create more jobs, and this will too. However, the rank and file of human beings that you call office vegetation also need a way to make money. We can't all be entrepreneurs. There will be mass layoffs coming - and already organizations like CNN have laid off a ton of journalists. This is as real as it gets.

We are in for a very, very rough ride. Dangerous, scary, exciting, and there's nothing on earth that will stop this from moving forward.

The singularity is a real thing, and it will happen.

I appreciate your skepticism and I would give anything for you to be right, because these things scare me.
AI, the pattern recognition kind we have now, can be used to detect and determine the obvious often with more acuity than humans but humans are still superior at exceptions to the obvious and well-known.
 
The scary thing about AI is how many people embrace it and trust the results without fully understanding what it can and cannot do.
Particularly how the underlying data that it is trained on and subsequently used with can have inconsistencies, missing data, systematic biases, ...

I have been analyzing Mental Health and Substance Use disorders for over thirty years as an Epidemiologist trained in Mathematics/Statistics.
I have been using Machine Learning and AI (along with other analytic methods) since the mid 1990s, so I have first hand knowledge on how ML and AI can lead to erroneous answers if you data is not carefully cleaned and checked for problems.

So,
My biggest concern is how ML and AI is often used without proper 'Real Intelligence' guiding it's use.
 
Thanks for sharing. I'm surprised it didn't surprise you. I really depends on the prompts you give it I suppose.

In any case, this will be interesting going forward.

Thanks,

Mike
I found it exceedingly dumb. I asked it a series of questions about audio transformers during which it recommended a book on audio transformers by Bill Whitlock. it even gave the ISBN number. I looked it up but the ISBN it had given me was for a completely different book - not even about transformers. So I told it this. it apologised and gave me a different ISBN. Long story short I did this about 10 times and each time it gave yet another wrong ISBN number. II tried once more and also asked it if it was absolutely sure this was the correct ISBN number and and it confirmed that it was but it still gave the ISBN of an unrelated book.

Not in the least impressive.

Cheers

Ian
 
I found it exceedingly dumb. I asked it a series of questions about audio transformers during which it recommended a book on audio transformers by Bill Whitlock. it even gave the ISBN number. I looked it up but the ISBN it had given me was for a completely different book - not even about transformers. So I told it this. it apologised and gave me a different ISBN. Long story short I did this about 10 times and each time it gave yet another wrong ISBN number. II tried once more and also asked it if it was absolutely sure this was the correct ISBN number and and it confirmed that it was but it still gave the ISBN of an unrelated book.

Not in the least impressive.

Cheers

Ian
I somewhat agree. But I wouldn't say it's exceedingly dumb. It really depends on what your asking. The response is usually put together very cleverly to 'not' give a definitive answer.

A couple months ago, I was talking with my nephew who heads one of Microsoft's ML/AI teams (they have two teams).
Microsoft owns a large portion of chatGPT and it runs on their Azure platform and he's very acquainted with the chatGPT algorithms.

My nephew said he doesn't really consider chatGPT as AI.
He said it's mostly web search, natural language processing and a tiny bit of AI.
He also said that in their 'Bing' implementation the answers returned are scrubbed to ensure that a 'definitive' answer is not given.
 
Back
Top