Below you find a transcript of my "conversation" with ChatGPT. I asked it to tell me how to create an 8 bar riser in Xfer Serum. Yes, I know how to do this – I don't need ChatGPT for this. I just wanted to test it and see if it has potential for sound design in general.

Spoiler: Chat transcript

ME: Tell me how to create an 8 bar riser in xfer serum
ChatGPT: Sure, here's a step-by-step guide to creating an 8 bar riser in Xfer Serum:

1. Open Xfer Serum and create a new blank patch.
2. In the oscillator section, select "Basic Shapes" from the dropdown menu for oscillator A. Choose a sawtooth waveform for oscillator A and turn up the wavetable position to about 75%.
3. Set the voices to "1" and detune the oscillator A slightly by adjusting the "Detune" knob to create some width and thickness to the sound.
4. Next, add an LFO to modulate the filter cutoff frequency. In the matrix section, select the first modulation slot and set the source to "LFO1". Set the destination to "Filter Cutoff" and adjust the amount to taste (around 50% is a good starting point).
5. In the LFO section, set LFO1 to a sine wave and adjust the rate to about 1/4 notes or whatever rhythm you want your riser to have.
6. Add some reverb to give the sound some space. In the effects section, click on the "FX1" tab and select "Reverb". Turn up the mix to about 50% and adjust the decay and size to taste.
7. Lastly, to create the riser effect, automate the filter cutoff frequency using the built-in automation features in your DAW. Create a 8-bar automation clip for the filter cutoff frequency and gradually increase it over the course of the 8 bars to create the rising effect.

That's it! You can adjust the settings to your liking and experiment with different waveforms, LFO shapes, and effects to create a unique sound that fits your track.

Me: Detuning when voices is set to 1 is redundant
CGPT: You are correct, detuning the oscillator when the "Voices" parameter is set to "1" would not have any effect on the sound since there are no additional voices to detune. Thank you for pointing that out.

To create a wider sound with just one voice, you could try adjusting the "Unison" parameter instead. Increasing the unison value will stack multiple detuned voices on top of each other, creating a thicker and wider sound. You could also experiment with other parameters like the "Spread" knob or using different waveforms to create more interesting timbres.

Me: What about pitch? A riser is about pitch change (over 8 bars in this case)
CGPT: You are absolutely right. Creating a pitch change over 8 bars is a crucial part of creating a riser. Here are the steps to add pitch automation in Xfer Serum:

1. In the matrix section, select the second modulation slot and set the source to "Envelope 1".
2. Set the destination to "Global Master Tune" and adjust the amount to taste.
3. In the envelope section, adjust the "Attack" parameter to be a little longer than 8 bars (the length of the riser). This will ensure that the pitch change is gradual over the entire length of the riser.
4. Create an automation clip for the envelope 1 attack time in your DAW and increase it gradually over the 8 bars to create the pitch rise effect.

You can experiment with different settings for the envelope and modulation amount to achieve the desired effect. You can also add additional pitch automation using other modulation sources like LFOs or macros to create more complex and dynamic pitch changes.

Me: I think using crs as destination makes more sense
CGPT: Yes, you're right. Using the "Course Pitch" destination instead of "Global Master Tune" would make more sense in this case. Here are the updated steps:

1. In the matrix section, select the second modulation slot and set the source to "Envelope 1".
2. Set the destination to "Crs" (short for Coarse Pitch) and adjust the amount to taste.
3. In the envelope section, adjust the "Attack" parameter to be a little longer than 8 bars (the length of the riser). This will ensure that the pitch change is gradual over the entire length of the riser.
4. Create an automation clip for the envelope 1 attack time in your DAW and increase it gradually over the 8 bars to create the pitch rise effect.

Using "Crse Pitch" as the destination will pitch shift the entire sound up or down by the specified amount, which is what we want to create the rising effect. Thank you for pointing that out!

Me: I don't need an automation clip since the envelope inside of serum already.
CGPT: Yes, you're right. Since Xfer Serum already has an envelope generator that can be used for pitch modulation, you can use that instead of creating an automation clip in your DAW. Here are the steps:

1. In the envelope section of Serum, click on the "ENV1" tab to open up the envelope editor.
2. Adjust the "Attack" parameter to be a little longer than 8 bars (the length of the riser). This will ensure that the pitch change is gradual over the entire length of the riser.
3. Set the envelope's sustain and release parameters to 0, so that the pitch change stops at the end of the 8 bars.
4. In the matrix section, select the second modulation slot and set the source to "Envelope 1".
5. Set the destination to "Crs" (short for Coarse Pitch) and adjust the amount to taste.
6. Preview the sound and adjust the envelope and modulation settings to achieve the desired pitch rise effect.

Using the envelope generator inside Serum to modulate the pitch will give you more precise control over the pitch rise effect. You can adjust the envelope shape, length, and modulation amount to create more complex and dynamic pitch changes.

As you can see, it is not as bad as expected but it probably won't nail it initially. And it seems like it doesn't have a clear "understanding" of my terms – like riser being associated with pitch. As we all know a riser could also be realized by automating the cutoff frequency of a LP filter when using white noise for example.

Have you done similar tests? What was the result?

 

It doesn't know shit about Harmony

 

Because of the breadth of information sources that the Chatgpt model has been taught, it makes it somewhat unpredictable how he will work with melodies and musical theories.
I have already spent a few days training my chat to match only the technical terms I defined and it worked well

 

He can generate melodies btw

 

Ask it: what melodic line comes to mind at this moment?

...and maybe he will create the exact song you think you were thinking at that moment but you couldn't spell it out.

 

score a simple quartet in the key of D - this is what it gave me. (should I call it it? maybe it has pronouns.)

Violin 1
D A B A G D D A
A D E D C# B A G
D A B A G D D A
A D E D C# B A G
D A B A G D D A
A D E D C# B A G
D A B A G D D A
A D E D C# B A G

Violin 2
D F# A F# E D D F#
A D F# D B A F# E
D F# A F# E D D F#
A D F# D B A F# E
D F# A F# E D D F#
A D F# D B A F# E
D F# A F# E D D F#
A D F# D B A F# E

Viola
D F# A F# E D D F#
A D F# D B A F# E
D F# A F# E D D F#
A D F# D B A F# E
D F# A F# E D D F#
A D F# D B A F# E
D F# A F# E D D F#
A D F# D B A F# E

Cello
D A B A G D D A
A D E D C# B A G
D A B A G D D A
A D E D C# B A G
D A B A G D D A
A D E D C# B A G
D A B A G D D A
A D E D C# B A G

This simple quartet consists of a repeating four-bar phrase that features a simple melody and harmonies in the key of D major. The melody is shared between the two violins, with the viola and cello providing accompanying harmonies and counterpoint.

Note that this is just one example of a simple quartet in the key of D, and there are many different ways to write for this ensemble. The specific style and musical elements will depend on the composer's preferences and the intended audience or performers.

 

when AI creates its own plugin I'll be impressed, right now its just a regurgitation of the internet, essentially ask it a question for a quick answer instead of finding the same answer yourself thru google. I am NOT sold on chatgpt for anything useful except answer questions I can find on the internet. The most use I got out of it was checking spl levels of drum pieces and seeing if certain mics I have would get damaged

 

AI Prompt Engineer is a new skill. Use reddit and youtube to learn the limits and get creative ideas for the AI tool you are using.

ChatGPT is the best tool to generate lyrics. It can review your lyrics and provide suggestions based on what you want to achieve. You can ask it to turn your lyrics into the style of Eminem. Or give it the technical details what you want to achieve, and it will do it for you. You can always ask it to edit the results until you get satisfied.

I think everybody should start find ways to use this in their daily life. It is a great tool.

 

Last edited by a moderator: Feb 22, 2023

Why plugins? when AI can create complete songs in very near future.

 

Sonic PI is programmable due the code system

you found some tutorials here.:

 

I can't seem to wonder, if the "Gods" that created us did the same thing to the first version of humans before we got updated / downgraded

 

I think AIVA is pretty impressive as AI composer tool

 

He easily optimizes code, creates presets for color correction, and translates text. It's a cool thing. Thinking about plus subscription.

 

I'm actually using it for mixing and getting great results. Of course it is a A.I., so if don't go especific it will tell some vague things. And you can actually ask the AI to improve the awnser if it doesn't satisfy you. Got some sick advices about mid range and stereo treatment with this thing.

 

Sounddesign ≠ Composing

 

Id rather it create plugins so there can be at least some life in the track, as the man said in the video, a human playing will sound better than what ai spits out... idk man, people who actually know how to make music are gonna find it easier to just play something instead of dicking around with some ai software. It seems like people who would use ai to write music have no concept of what they want to hear or have no idea how to take shit they hear in their head and turn it into a song, like I can and other people with experience can - thats why I would prefer a plugin, or like the keyboard he mentions in the video (i play a melody and ai spits out a bunch of different cord progressions I can put under it).

 

what did it say about the mid range - like in what aspect?

 

To my mind using ChatGPT outside of a basic search function is contributing to the dumbing down of humanity and will water down and eliminate creativity in the short long term. From what I have witnessed so far is that it is a tool to think up ideas in lieu of one creating from personal experience, emotion gathered through living and breathing a life and connections groked or imagined. ChatGPT makes it a thing to be disengenuous while allowing the user to take credit for the end result in a way to me likening it to a form of plagerism if used to concoct songs, stories, poetry or any other endeavor which a person can take credit for without letting on to his/HER artificial intelligent collaborator. This AI thing is a commercial endeavor to make as much money as is "humanly" possible (which is the sole goal of any corporation and the fiduciary responsibility of the board of directors to their stockholders...everything else like treating employees fairly or the safety of products is secondary to the mission) . Since it is now possible to keep tabs on every instance of every communication with ChatGPT it will only be a little while until whoever owns the corporation that eventually owns it will begin extracting money as co-creator of any commercial success that a user enjoys from using it on this way. Like everything else this "tool" will learn and become better and better until there will a very narrow margin between what a fantastic artist creates and what ChatGPT will deliver. Remember this...like so much of the garbage foisted upon us now by those who control the media and hold great sway and power concerning propaganda and doing their best in controlling their narrative, soon enough it will be simply cost effective and preferred to cut the human out of the equation entirely in order to reap a crop of gold in any field as they strew their genetically modified toxicity. Amazon is getting rid of delivery people in favor of unmanned trucks and drones, the creative arts will seemingly soon follow. Siding with your jailer is a kind of insanity and is exactly like the symptoms of The Stockholm Syndrom. Don't be the frog in the pot of water that doesn't realize that the water is slowly getting hot and will soon cook that frog to death. Realize that the water is already boiling and the moment that you touch the water skip right out to survive another day.

I implore you.