You can create a custom ChatGPT for each synthesizer. For example, by uploading the Iridium manual, you create a unique GPT and ask for the creation of sounds. Of course, it will be just a matter of time until we see VSTs with a language model built in. For now, this is an interesting workaround.
If you use Perplexity - particularly the subscription version (most up-to-date AIs) - no uploading is required, because it can access *current* web content, unlike Chat-GPT alone which accesses archived material. Very useful, being able to incorporate manufacturers data etc.
This was great to watch, Martin. My take on AI for any creative use: photography, art, music, writing.... it's a tool. If someone is making woodwork and they buy a new clever tool that lets them cut circles (for example) in 1/2 time, nobody says "oh that's not craft, that's someone else doing it for you..." so why is AI any different. If the computer is MAKING music without you putting any of your own creativity into it, then it's not your work. But, if you're using it for set up, and seeding ideas you might not have had before, and then it's all your work after that, I see no problem. People need to think it through.
this is great! i will make custom gpt's for my synths and start some sounddesign sessions. this helps a lot moving forward from just selecting presets. thank you martin
It's a little but like choosing a preset but being guided in how to create it. Like this you get a better understanding and are enabled to tweak and use it creatively the way you want!
Martin, a possible new idea... ask ChatGPT " How can I get my cat to play a 3Hr Home Concert so I can relax with a cup of tea?" Thanks for explaining what the capabilities are and good to see it can work out
I agree with your comment at the end of the video about using creative tools (ai or randomization) to help users learn their synthesizers. I had a patch randomization function on my rack Waldorf Microwave 2, but I never really found it useful as an exploratory tool because of the limited knobs/buttons on the device to tweak the randomized patch. However when I got an ASM Hydrasynth, I found patch randomization to be really, really useful tool in learning how to program the synth, due to the intelligent layout of the synth inviting quick editing of the patches. I recently enjoyed using the same random functionality when pairing an Ensoniq Fizmo with editing software that has a really fantastic randomization function, it opened up the possibilities with the synth that I probably would have never understood without the software assistance. Also, I love that Neptun knows that when you start a video, he needs to make an immediate appearance.
Great to see your approach. I've been using it on and off to help me stay on track when creating my own sounds. Sometimes I forget about certain destinations when modulating patches, for example, and using my GPT helps me ensure that if I leave something out is a design choice and not that I simply forgot. I also ask for suggestions for classic synth sounds, and try to replicate that with the synths I have.
Been doing the same thing with PhasePlant (modular synth), Gaea (a terrain generator program I was learning how to use it) and using Phyton without knowing anything else than the name, to create Scripts for the FL Studio Piano Roll... ChatGPT is great!
Hi Martin, yes, I agree with you. AI can be helpful for beginners. I have also different experiences with AI (gemini, udio). Very often it helps, sometimes the anserws are incorrect, and the generated music a la Berlin style is fine, but a little bit boring. In general, we should use our brain and think about the answers ... (this is always valid). Best Frank
@MartinStuertzer There is a copy/paste function, of sorts (the button to the right of the operator level setting). It's a bit of a blunt tool, as it copies *every* setting of the source operator and applies it to the target. But better than nothing, I suppose. Also, a tip for the people wary of tangling with the DX7: @One Man And His Songs has a 10-episode walkthrough/tutorial on it. The episodes on envelopes were eye-opening. Highly recommended.
Very funny video , thanx ! :) @0:12:00 : Yes it doesnt know because it s a TCN (Text Classification Network, if I m correct) :D . So this model has been trained on web publications released from the last 20-30 years then render, aggregate, classify those and that s it. :) It could be very usefull for "lazy" beginers who do not want to read the MANUALS and experiment without making any mistakes.... , but pro : no way ..... :D :D . But very funny topic indeed !!
Wouldn't it be easier if ChatGPT created a patch file which you download into your desired synth? Or upcoming synths with LLMs inside dialing in your desired sounds?
Great video! And a logical conclusion. AI is just an advanced Google of the modern era. We will learn to use it effectively and creatively. Its just a mater ot time...
Hallo Martin - super Idee, ich habe vor ca. 10 Monaten den Osmose gegönnt - vielleicht sollte ich ChatGPT mal in Richtung Sounddesign mit der doch komplexen Haken Audio Engine traktieren? FM Synthese und subtraktive oder Wavetable Synthese sind ja vergleichsweise "einfache" Soundgeneratoren. 🙂 Gruss aus Hattingen
Haken ist aber wirklich der Endgegner. Ich hatte mal den Mini Continuum hier und bin schon daran gescheitert, den Editor auf meinem Computer mit dem Gerät so zu verbinden, dass nicht alle Presets gelöscht werden.
interesting. Since we all hate ai in music prod (ahem), this is a clever take on how to use AI this time more like a learning tool. I allow this 'p ...becomes a teacher in a way... well at least i mean, im at the prophet 5 section for now....should i watch the rest without fear ?
a program randomizer and some tweeking can get you a long way .. A.i. is going to be over used i personally find it exhausting to go thru all the media you can create with ai in a few clicks and afew prompts it is better just to learn your gear and make your own music
Why learn when you can carry water for tech billionaires instead, right? Who wouldn't want to insert an unhelpful algorithm-controlled step or steps into his or her thinking process?
Ssssssssssssshut up! ;) Ich habe meine Vocal Chain etwas überarbeitet und jetzt auch einen De-Esser drin. Ab dem nächsten Video sollte der Ton etwas besser sein.
Marthin says "be nice" then starts his rant by calling names "haters" and making accusations "you probably didn't watch the whole video". Not only is that the definition of passive aggressive but also committing multiple logical fallacies by insulting people who might have any criticism. I and many people have nuanced opinions and see both sides of the issue. But in this context - AI is the opposite of Zen. AI is not about being in the moment, enjoying the process of life, it's about results. I feel sorry for people who are completely missing out on the essential part of experience that makes being human valuable - instead trading it in for being a tool for AI to create a product of decreasing value. Maybe you would "never have come up with this on your own" as you say Martin, because you didn't take the time to go and explore new things and new approaches but rather instead just put in a prompt. Sorry man you lost me on this one.
It was hardly a rant and he calmly and patiently explained as best he could his reasons for making the video, not a diatribe on the state and effect of AI overall. Explaining reasons and still being reasonable is especially difficult in a language not native to you
You are so off-base here. The only rant is yours. He wasn’t passive aggressive but you are aggressive and offensive. I would say more but I’ve fed the troll too much already. It’s not worth it. I hope your night, your weekend, your LIFE, all get better and soon. Good grief.
I had to watch the video twice to see if I missed something passive aggressive, only to realise that the op was the one who missed the point completely.
You can create a custom ChatGPT for each synthesizer. For example, by uploading the Iridium manual, you create a unique GPT and ask for the creation of sounds. Of course, it will be just a matter of time until we see VSTs with a language model built in. For now, this is an interesting workaround.
If you use Perplexity - particularly the subscription version (most up-to-date AIs) - no uploading is required, because it can access *current* web content, unlike Chat-GPT alone which accesses archived material. Very useful, being able to incorporate manufacturers data etc.
Thank you for sharing all this, it's a source of knowledge.
Thank you for showing how ChatGPT can help us create a synth sound. It's not just generating a sound, it's rather a guide to learn sound design.
Exactly! That is, what I like about it. You can still shape the sound to your needs.
The velocity sensitivity is in the overview pane (DX-7) .. to the right of cutoff and left of pan 🙂
Thank you!
This was great to watch, Martin. My take on AI for any creative use: photography, art, music, writing.... it's a tool. If someone is making woodwork and they buy a new clever tool that lets them cut circles (for example) in 1/2 time, nobody says "oh that's not craft, that's someone else doing it for you..." so why is AI any different. If the computer is MAKING music without you putting any of your own creativity into it, then it's not your work. But, if you're using it for set up, and seeding ideas you might not have had before, and then it's all your work after that, I see no problem. People need to think it through.
I really enjoyed this video. I never would’ve thought to use ChatGPT to help get sound design ideas out of my head
this is great! i will make custom gpt's for my synths and start some sounddesign sessions. this helps a lot moving forward from just selecting presets. thank you martin
It's a little but like choosing a preset but being guided in how to create it. Like this you get a better understanding and are enabled to tweak and use it creatively the way you want!
Martin, a possible new idea... ask ChatGPT " How can I get my cat to play a 3Hr Home Concert so I can relax with a cup of tea?" Thanks for explaining what the capabilities are and good to see it can work out
A very good idea! Neptun will also be happy to have less work to do.
I like that 😆
I agree with your comment at the end of the video about using creative tools (ai or randomization) to help users learn their synthesizers. I had a patch randomization function on my rack Waldorf Microwave 2, but I never really found it useful as an exploratory tool because of the limited knobs/buttons on the device to tweak the randomized patch. However when I got an ASM Hydrasynth, I found patch randomization to be really, really useful tool in learning how to program the synth, due to the intelligent layout of the synth inviting quick editing of the patches. I recently enjoyed using the same random functionality when pairing an Ensoniq Fizmo with editing software that has a really fantastic randomization function, it opened up the possibilities with the synth that I probably would have never understood without the software assistance.
Also, I love that Neptun knows that when you start a video, he needs to make an immediate appearance.
Great to see your approach. I've been using it on and off to help me stay on track when creating my own sounds. Sometimes I forget about certain destinations when modulating patches, for example, and using my GPT helps me ensure that if I leave something out is a design choice and not that I simply forgot. I also ask for suggestions for classic synth sounds, and try to replicate that with the synths I have.
Awesome video!! Thanks for doing this….DX7 usually scares me but with GPT’s help you made it seem very approachable.
Neptun .... nich gegen's Mikrophone ... echt mal. Danke mein Guter, für die tolle Idee mit Chat GPT. Großartig!
Das hat mir sehr gut gefallen. Interessanter Einblick ins Klangdesign mithilfe eines LLMs. Danke! 👍
It's nice to know that AI isn't out to hurt anybody. Violence dank 😊
Thank You Martin!! I prefer learning from you and Chris!
Any day!
My husband has had lengthy conversations with GPT 4-o over hardware synth designs too . Even with cost breakdowns.
Been doing the same thing with PhasePlant (modular synth), Gaea (a terrain generator program I was learning how to use it) and using Phyton without knowing anything else than the name, to create Scripts for the FL Studio Piano Roll... ChatGPT is great!
Hi Martin, yes, I agree with you. AI can be helpful for beginners. I have also different experiences with AI (gemini, udio). Very often it helps, sometimes the anserws are incorrect, and the generated music a la Berlin style is fine, but a little bit boring.
In general, we should use our brain and think about the answers ... (this is always valid).
Best Frank
I can't wait for the "StürtzerGPT" so it trains me!
Fun! Did this exact same thing with the hydrasynth not so long ago :)
This is good news, I forgot how to use my nord g2 editor as it's been so long i've used it, now i can just ask chat gpt to guide me through it :D
@MartinStuertzer There is a copy/paste function, of sorts (the button to the right of the operator level setting). It's a bit of a blunt tool, as it copies *every* setting of the source operator and applies it to the target. But better than nothing, I suppose.
Also, a tip for the people wary of tangling with the DX7: @One Man And His Songs has a 10-episode walkthrough/tutorial on it. The episodes on envelopes were eye-opening. Highly recommended.
finally someone thats actually useing AI properly. yes we can all learn from it..
A sensible use of AI in music production. 👍
Very funny video , thanx ! :) @0:12:00 : Yes it doesnt know because it s a TCN (Text Classification Network, if I m correct) :D . So this model has been trained on web publications released from the last 20-30 years then render, aggregate, classify those and that s it. :) It could be very usefull for "lazy" beginers who do not want to read the MANUALS and experiment without making any mistakes.... , but pro : no way ..... :D :D . But very funny topic indeed !!
Very impressive, what modern software can do! And as usual: AI generated contents need a little supervision/control. But if one keeps that in mind...
This is both funny and interesting.. :-)
Wouldn't it be easier if ChatGPT created a patch file which you download into your desired synth? Or upcoming synths with LLMs inside dialing in your desired sounds?
Sure, if you just want the patch. I like the idea to follow the guide step by step to understand the patch and also change it to my own taste.
please make ambient with serume
♥
Great video! And a logical conclusion. AI is just an advanced Google of the modern era. We will learn to use it effectively and creatively. Its just a mater ot time...
DESSSEER ! great video btw
Sir please can you tell me the last ended background sound name of the stream was what?
Great idea! Thanks for trying this out and sharing the experience.
ありがとうございます。
お願いします！
Really nice! LLM is a great tool to learn new things. I have become a bit more critical of AI art which looks all the same.
What are the best advantages of this Waldorf synth?
Why choose this over Vital?
Hello, What DAW you used?
Hallo Martin - super Idee, ich habe vor ca. 10 Monaten den Osmose gegönnt - vielleicht sollte ich ChatGPT mal in Richtung Sounddesign mit der doch komplexen Haken Audio Engine traktieren?
FM Synthese und subtraktive oder Wavetable Synthese sind ja vergleichsweise "einfache" Soundgeneratoren. 🙂
Gruss aus Hattingen
Haken ist aber wirklich der Endgegner. Ich hatte mal den Mini Continuum hier und bin schon daran gescheitert, den Editor auf meinem Computer mit dem Gerät so zu verbinden, dass nicht alle Presets gelöscht werden.
@@MartinStuertzer Naja, wie soll AI denn lernen, wenn nicht mit Herausforderungen 😉
💖
Interesting idea. I really like the pad on the DX7.
Sound design with Cat GPT... thanks for the video
interesting. Since we all hate ai in music prod (ahem), this is a clever take on how to use AI this time more like a learning tool. I allow this 'p ...becomes a teacher in a way... well at least i mean, im at the prophet 5 section for now....should i watch the rest without fear ?
the DX7 sound is rather good
I would say "thank you", but I just followed the steps in the manual.
a program randomizer and some tweeking can get you a long way .. A.i. is going to be over used
i personally find it exhausting to go thru all the media you can create with ai in a few clicks and afew prompts
it is better just to learn your gear and make your own music
Why learn when you can carry water for tech billionaires instead, right?
Who wouldn't want to insert an unhelpful algorithm-controlled step or steps into his or her thinking process?
Interessante Idee, versuche ich auch mal demnächst.....vielen Dank!
hm. immer text lesen .... oder gleich alles mit AI machen lassen ... nur dann ist "Musik-machen" hinfällig, so wird es kommen
No hating aber den Zischlauten deiner Stimme täte ein Deesser ganz gut
Ssssssssssssshut up! ;) Ich habe meine Vocal Chain etwas überarbeitet und jetzt auch einen De-Esser drin. Ab dem nächsten Video sollte der Ton etwas besser sein.
This is still tech for lazy young lawyers to me. 😂 I don’t appreciate it for music.
Marthin says "be nice" then starts his rant by calling names "haters" and making accusations "you probably didn't watch the whole video". Not only is that the definition of passive aggressive but also committing multiple logical fallacies by insulting people who might have any criticism. I and many people have nuanced opinions and see both sides of the issue. But in this context - AI is the opposite of Zen. AI is not about being in the moment, enjoying the process of life, it's about results. I feel sorry for people who are completely missing out on the essential part of experience that makes being human valuable - instead trading it in for being a tool for AI to create a product of decreasing value.
Maybe you would "never have come up with this on your own" as you say Martin, because you didn't take the time to go and explore new things and new approaches but rather instead just put in a prompt. Sorry man you lost me on this one.
It was hardly a rant and he calmly and patiently explained as best he could his reasons for making the video, not a diatribe on the state and effect of AI overall. Explaining reasons and still being reasonable is especially difficult in a language not native to you
who are you to judge what experiences others want to experience? wooly thinking..... and ZEN is what ever you want it be, without worrying......
You are so off-base here. The only rant is yours. He wasn’t passive aggressive but you are aggressive and offensive. I would say more but I’ve fed the troll too much already. It’s not worth it. I hope your night, your weekend, your LIFE, all get better and soon. Good grief.
I had to watch the video twice to see if I missed something passive aggressive, only to realise that the op was the one who missed the point completely.
Wow. Cool down man. I think you might have come in a bit hot there.