So, I got into a 'conversation' with CoPilot about Tonex and it asked me if I wanted it to guide me to make a particular sound - I'd read that the Satriani presets were good for jazz guitar. I said ok and it walked me through the parameters, saying why each setting worked for that particular tone. I tried it again for Amplitube for a Tony Iommi sound and again, it was comprehensive. I haven't actually tried dialling these in but I suspect they will be in the ball park - too many variables for a real 1 to 1 sound. I guess this will work for any hardware device, synth or guitar effects processor and any plugin, for anything. Seems like a great starting off point instead of mindlessly scrolling through thousands of presets.

 

yea you can use something like open-source Stable Audio and generate any sounds/samples
https://huggingface.co/stabilityai/stable-audio-open-1.0

 

yes, AI is great at this sort of thing. I use chatgpt for synth recipes, as well as tone recipes. It knows everything there is to know about recording techniques. You can say something like " i need you to tell me the exact methods the Beatles used to make Revolver. Be as thorough as possible, include what equipment was used in all parts of the signal chain, what their mic-ing techniques were, how was the room treated, any settings for hardware or instruments, etc. Only include factual information, do not speculate or improvise."

 

you can upload a text of your entire plugin inventory to chatgpt, save it, and then get it to do some pretty complex stuff like this, but with channel strip chains, ideas for routing, etc. It is definitely worth giving it a try, no matter how well you already know your stuff. I had not thought about using it for synths, but i'm going to give that a try now too. I wonder if it can create text patch sheets from pictures taken of analog synths.

 

Spoiler: Source: Grok KI

That’s a solid observation! AI like Copilot (or me) can break down the parameters for specific tones, like Satriani’s jazz guitar or Iommi’s heavy doom sound, by analyzing the characteristics of those tones and mapping them to the controls of tools like ToneX or AmpliTube. It’s a smart starting point—way more efficient than aimlessly scrolling through presets. For hardware like synths or guitar effects processors, or even plugins, the same logic applies: we can suggest settings based on the desired sound profile, factoring in things like gain, EQ, modulation, or resonance that define the artist’s tone.

You’re right that nailing an exact 1:1 sound is tricky due to variables—gear, playing style, signal chain, even room acoustics. But getting in the ballpark with guided parameters is a huge time-saver. If you want, I can walk you through settings for a specific tone on ToneX, AmpliTube, or any other gear—say, a particular artist or genre. For example, Satriani’s jazz tone on ToneX might lean on a clean amp model like a Fender Deluxe, with moderate gain, boosted mids, and a touch of reverb for airiness. Want me to dive into a specific setup for you to try? Or maybe explore another artist’s sound, like Iommi’s Sabbath crunch?

 

excellent idea to upload your plugin list. Its usually pretty good with synths. Id say 80% of the time its given me a great recipe. It definitely understands synthesis and advanced techniques, but it can sometimes be hard explaining what it is your after. I have uploaded audio and gotten synth recipes from that as well. its hit or miss.

 

I feed the prompts with my template signal flow, describing it precisely. Every time i get sick of some sound i get a huge amount of help from DeepSeek (not GPT) because with DeepThink i can access the reference material that always ends up on a good read on trusted blogs like SoundOnSound. They even recommend underground Nebula Libs and Airwindows plugins, not commercial or ad heavy plugins like GPT.

You can also send Michael Brauer template on AI and start tweaking from there. But be careful: sometimes it alucinates, but you can always correct the bot if you know the flaws on the answer.

Spoiler: My personal experience

I feed the prompts with my template and especific sections of MB workflow and ended up with a really cool 70's console, instead of a SSL 9000 console, which suited better for the sound that i want

 

Exactly where I first got the idea. We have so many different plugins that are nearly interchangeable in your template. I wanted one for each genre I mix. I do not do "set and forget" with anything really, it's just a huge time saver to have all my buses setup, each chain already figured out the main direction I want them to go in the template. Some things are just there as placeholders. I know I want a clipper or a compressor in a certain insert slot no matter what, as an example. So maybe a vanilla Pro-C2 instance is sitting there. Doesn't mean I can't swap it with a UAD, Acustica, or whatever other plugin that will sound better. Everything I know I will want is already there, but disabled and not eating resources.

 

Hotone has already integrated AI to its software. You type a quick prompt and the AI will generate a bunch of presets which you can then directly test on your multi-fx unit and download.
I tried it and it works pretty well as a starting point. It even came up with pretty decent tones for very big names like (rolling stones, metallica and the likes) though I doubt you'd get anything (maybe!) if you were to type Tony Iommi.
It's not really the goal either at this stage (it's in beta) but this is definitely a feature we'll see more and more in the very near future.

 

I havent spent any time with DeepSeek, so im not sure if this applies, but with ChatGpt you can tell it to behave in certain ways, and save those settings as a new GPT for use. I have sets of rules that keep it from hallucinating, its been a game changer. LMK if DeepSeek has similar and i can share the rules. Its a few paragraphs of text that ive spent weeks researching and optimizing.

 

I leave everything enabled so i can expect the amount of CPU every project will use on average and i kinda hate linearity. But even if everything is already online, i abuse the bypass button to check if it's helping, not making any difference or making it worst. No heavy processing, eq presets, anything like that; just basic analog signal flow to make the DAW react like a mixing console. If i think x plugin will be the best i will use it forever (like Vertigo VSC-3 on drums; neutral, groovy and with some creamy VCA vibe with slower releases). Some will never abandon their Distressors or API-2500 on the same material.

And IMO it's hard to find a chain by yourself that will interact well in-between processings: there's when you can ask help from AI. I could find some crazy chains and crazy advanced stuff that fulfilled my need for real.

It's the same concept, so it will be welcomed! I find DeepSeek more scientifical than GPT (because of the DeepThink stuff for referencing) but they work the same way. Any help to prevent it will be great

 

Blimey, I'm so glad I read this post. Yesterday I spent hours writing Reabank files, the articulation manager in Reaper. Not complicated but easy to make mistakes and what with debugging takes a lot of time. I just registered with Deepseek and it just produced in seconds what it took me hours to do myself. Not only that but kept coming back with suggestions about taking it further or do I want a script to automate the whole process etc etc.

 

And can it create a synth preset based on the submitted sound? Let's say we take the synth intro from ABBA Gimme! Gimme! Gimme! and an existing synth plugin (say Silent1 or Massive) and ask it to create the matching preset(s)?

 

You don't even need to submit the sound!

Spoiler: CoPilot

CoPilot:

The iconic intro synth from ABBA’s “Gimme! Gimme! Gimme!” is a bright, flute-like lead with a slightly nasal, vintage character—originally played on an ARP Odyssey. You can get surprisingly close in Serum with a bit of clever layering and filter shaping.

Here’s a Serum patch blueprint to get you started:

ABBA “Gimme! Gimme! Gimme!” Intro Synth in Serum
Oscillators

• Osc A: Saw wave, 1 octave up, level ~70%
• Osc B: Pulse wave (use Basic Shapes, shape 3), Pulse Width ~90%, level ~60%, detune slightly
• Unison: 2–3 voices on Osc A for subtle width

️ Filter

• Type: MG Low 12
• Cutoff: ~8.5 kHz
• Resonance: ~20%
• Drive: ~10–15% for warmth
• Route both oscillators through the filter

Envelope 1 (Amp)

• Attack: 0 ms
• Decay: ~400 ms
• Sustain: 100%
• Release: ~150 ms

️ Envelope 2 (Modulation)

• Assign to filter cutoff
• Attack: ~10 ms
• Decay: ~300 ms
• Sustain: 0%
• Amount: ~30–40 (mod wheel can control this too)

FX Section

• Chorus: Mix ~30%, Rate ~0.2 Hz, Depth ~40%
• Delay: Ping-pong, 1/8 dotted, Mix ~25%
• Reverb: Plate or Hall, Size ~40%, Mix ~30%
• EQ: Boost highs slightly, cut low mids

Bonus Tips

• Play in mono legato mode with a touch of portamento (~30 ms) for that gliding feel.
• Add a subtle vibrato via LFO 1 to pitch (rate ~5 Hz, depth ~10 cents).
• If you want to layer it with a string pad underneath (like the original), try blending with a soft saw pad in your DAW.

If you’d like, I can export this as a .fxp preset or walk you through building it step-by-step in Serum. Or we could even recreate the full intro stack with strings and bass. Want to go deeper?

 

Last edited by a moderator: Jun 29, 2025 at 9:29 PM

In my experience, Copilot or ChatGPT cannot directly generate serum (or any other vst) presets on their own format.



When I asked Copilot to do this I got this answer:
"
can you recreate synth from ABBA’s “Gimme! Gimme! Gimme! on serum vst preset and export is as .fxp ?

I can absolutely help you recreate the iconic synth sound from ABBA’s “Gimme! Gimme! Gimme!” using Serum's tools, but I can't directly export or send you a .fxp file. That said, I can walk you through the sound design process so you can build it yourself—and we can get really close to that lush, dramatic ‘70s synth vibe.

... ... ...

why you can't directly export or send me .fxp?

That's a great question—honestly, I wish I could! The reason I can’t generate or export .fxp files is simply because I don’t have direct access to your file system or external plugin formats. I'm designed to assist with creative guidance, walk you through the process step by step, and even give you detailed settings to replicate—but the actual act of generating a plugin-specific file like .fxp goes beyond what I’m built to do.

Think of me like your sound design co-pilot sitting in the passenger seat—I can read the map, plan the route, and even point out some cool detours, but you're the one behind the wheel tweaking the knobs and clicking Save.

If you want, I can format the settings in a way that’s easy to enter into Serum. Or if you're working on a track and want to craft several presets for it, I can help you build a whole bank of sounds to suit the vibe. Want to give that a shot?


which AI can do it?

At the moment, there aren’t any mainstream AI models that can directly create and export .fxp files for Serum or other synth plugins inside your DAW. Even if an AI could theoretically generate the correct data, it would still need to integrate with your DAW environment and obey plugin file structures—a pretty complex task requiring specific software permissions and format handling.

 

ChatGPT can create source codes you can save as presets. At the moment for Diva and Hive2 (h2p-format)

 

^ that is all they can do with mixing templates too. They cannot generate project files you can actually load.

Here is an example drum bus chain I just asked it to make from my plugin inventory:

It's still more than I would type out or tell someone. I'm not interested in the settings in this example, as I said they would just be starting points anyway. If you need it to actually spit out a file to set this amount of basic parameters, that is beyond lazy. A synth patch would be the same thing, and for analog synths even better, if they do not have any other recall options beyond taking pictures.


Logic Pro House Music Drum Bus Template (Stereo, 48kHz, 24-bit)

BUS: Drum Bus (Bus 2) – Kick, Snare, Hats, Loops
Color: Red
Track Name: DRUM BUS

Plugin Chain:
1. Channel EQ (Logic Pro X)
- High-pass @ 25Hz (12dB/oct)
- Cut @ 250Hz (-2.5dB, Q=1.0)
- Presence boost @ 6kHz (+1.5dB, Q=1.0)

2. SSL 4000 G Bus Compressor or bx_glue
- Ratio: 4:1
- Attack: 30ms
- Release: Fast or Auto
- GR: 2–4 dB
- Mix: 100%

3. KRAФTUR (Soundtheory)
- Drive: +4 dB
- Offset: 3 dB
- Knee: 1.5 dB
- Band Shifts: Mid +3 dB, High +1.5 dB
- Blend: 70% Single Band
- Clip Ceiling: ON

4. KClip 3 (Kazrog) – Mode: CRISP
- Input Gain: +4 dB
- Ceiling: -0.1 dB
- Mix: 100%
- Oversampling: 8x
- Multiband: OFF

5. Limiter (Optional – Logic Adaptive Limiter)
- Output Ceiling: -0.1 dB
- Lookahead: ON

Routing:
- All drum tracks (kick, snare, hats, loops) output to Bus 2.
- Bus 2 routes to Stereo Output.

Notes:
- Auto Gain ON in bx_glue or KClip for easy A/B.
- bx_glue Saturation can be set to 10–15% for analog feel.
- M/S clip optional in bx_clipper if used instead.

 

I want to recreate the song "Gimme Giime Gimme" by the band Abba for the Xfer Serum VSTi. Do you have the data for it?

Spoiler: Source: Grok KI (German)

Um den ikonischen Lead-Synthesizer-Sound von ABBAs „Gimme! Gimme! Gimme! (A Man After Midnight)“ mit Xfer Serum nachzubauen, kannst du die folgenden Schritte und Einstellungen als Leitfaden verwenden. Der Sound basiert auf einem ARP Odyssey, der mit einer Yamaha SS-30 String-Machine gedoppelt wurde, und hat einen hellen, flötenartigen Charakter, der mit Streichern unterlegt ist. Hier ist ein allgemeiner Ansatz, basierend auf verfügbaren Informationen und typischen Synthesizer-Techniken:

Einstellungen für Xfer Serum

1. Oszillatoren:
   • Oszillator A: Wähle eine Sägezahn-Wellenform (Saw), da diese dem hellen, vollen Klang des ARP Odyssey nahekommt. Aktiviere Unison mit 4-6 Stimmen und einem leichten Detune (ca. 0.05-0.10), um den Klang breiter zu machen.
   • Oszillator B: Verwende eine Dreieck-Wellenform (Triangle), um den flötenartigen, weicheren Aspekt des Sounds zu erzeugen. Reduziere die Lautstärke von Oszillator B leicht, damit er den Sägezahn unterstützt, ohne zu dominieren.
   • Stelle die Oktave von Oszillator B eventuell eine Oktave höher, um die höheren, flötenartigen Obertöne zu betonen.
2. Filter:
   • Verwende einen Tiefpassfilter (Lowpass) mit einer Cutoff-Frequenz von etwa 1-2 kHz, um den Klang weich und warm zu halten, aber dennoch die Helligkeit zu bewahren.
   • Resonance: Halte die Resonanz niedrig bis mittel (0.2-0.4), um subtile Obertöne zu betonen, ohne den Klang zu „schreien“.
   • Aktiviere eine leichte Filterhüllkurve (Envelope), um die Cutoff-Frequenz dynamisch zu modulieren. Stelle die Attack-Zeit auf kurz (ca. 10-20 ms) und Decay auf mittel (ca. 500 ms), Sustain auf 50-70 % und Release auf 100-200 ms.
3. Hüllkurven (Envelopes):
   • Amplitudenhüllkurve: Kurze Attack-Zeit (0-10 ms), mittlerer Decay (ca. 500 ms), Sustain auf 70-80 %, und ein kurzes Release (ca. 100-200 ms), um den Klang knackig, aber nicht zu abrupt zu gestalten.
   • Weise die Hüllkurve auch dem Filter-Cutoff zu, um den Klang dynamisch zu öffnen.
4. Modulation:
   • LFO: Weise ein LFO mit einer langsamen Rate (0.5-1 Hz) und einer Dreieck-Wellenform dem Filter-Cutoff oder der Panorama-Position zu, um eine leichte Bewegung im Klang zu erzeugen, die den Streicher-Layer im Original simuliert.
   • Optional: Füge eine subtile Modulation der Tonhöhe (Pitch) hinzu, um den Klang lebendiger zu machen (sehr geringe Tiefe, z. B. 0.05-0.1).
5. Effekte:
   • Chorus: Aktiviere den integrierten Chorus in Serum, um den Klang breiter und „vintage“ zu machen. Stelle die Rate auf langsam (0.3-0.5 Hz) und die Tiefe auf mittel.
   • Reverb: Füge einen leichten Hall hinzu, um den Klang räumlicher zu gestalten. Halte die Decay-Zeit kurz (1-2 Sekunden) und den Mix niedrig (10-20 %), um den Disco-Charakter zu bewahren.
   • Delay: Ein leichtes Stereo-Delay (1/8 oder 1/16 Noten) mit geringem Feedback (20-30 %) kann dem Klang mehr Tiefe verleihen, ohne ihn zu überladen.
6. Layering für Streicher-Effekt:
   • Um die Yamaha SS-30 String-Machine zu simulieren, füge einen zweiten Patch in Serum hinzu oder nutze ein separates String-VST. Wähle eine sanfte Streicher-Wellenform (z. B. „Analog Strings“ in Serum) und halte die Attack-Zeit etwas länger (50-100 ms), um den weichen Streichercharakter zu erzeugen.
   • Mische diesen Layer leise unter den Lead-Sound, damit er den Hauptsynthesizer unterstützt, ohne ihn zu überdecken.

Zusätzliche Hinweise

• Referenzen: Laut Quellen wurde der Lead-Sound hauptsächlich mit einem ARP Odyssey erzeugt, gedoppelt mit einer Yamaha SS-30 String-Machine. Der Bass ist eine Kombination aus Minimoog und Bassgitarre. Diese Informationen können dir helfen, den Klang authentischer nachzubauen.
• Tutorial und Presets: Es gibt ein YouTube-Tutorial, das den Lead-Sound speziell für Serum nachbaut, inklusive eines kostenlosen Presets: YouTube - How to Make ABBA "Gimme Gimme Gimme" Lead Synth. Dieses Tutorial zeigt detaillierte Einstellungen und kann eine gute Grundlage sein.
• MIDI-Daten: Falls du MIDI-Daten für die Melodie benötigst, gibt es professionelle MIDI-Dateien, z. B. bei midi.com.au, die die Melodie von „Gimme! Gimme! Gimme!“ enthalten. Diese können in deiner DAW importiert werden, um die Noten genau nachzuspielen.

Einschränkungen
Ich habe keine direkten Preset-Daten (z. B. eine .fxp-Datei für Serum), da solche spezifischen Dateien oft urheberrechtlich geschützt sind oder von Drittanbietern bereitgestellt werden. Du kannst jedoch die oben genannten Einstellungen manuell in Serum einstellen oder das Preset aus dem erwähnten YouTube-Tutorial herunterladen. Wenn du Schwierigkeiten hast, den Sound exakt hinzubekommen, empfehle ich, das Tutorial-Video zu verfolgen oder ein vorgefertigtes Preset von Seiten wie syntorial.com zu verwenden, die speziell für diesen Song erstellte Presets anbieten.




Falls du weitere Details oder Hilfe bei spezifischen Serum-Einstellungen brauchst, lass es mich wissen!

Spoiler: Source: Grok KI (English)

To recreate the iconic lead synthesizer sound from ABBA’s “Gimme! Gimme! Gimme! (A Man After Midnight)” using Xfer Serum, you can follow the guidelines below. The sound is based on an ARP Odyssey, doubled with a Yamaha SS-30 String Machine, delivering a bright, flute-like lead with a string layer. Here’s a step-by-step approach based on available information and typical synth design techniques:

Xfer Serum Settings

1. Oscillators:
   • Oscillator A: Select a Sawtooth Waveform to mimic the bright, rich tone of the ARP Odyssey. Enable Unison with 4-6 voices and a slight Detune (0.05-0.10) for width.
   • Oscillator B: Use a Triangle Waveform to capture the softer, flute-like quality. Lower Oscillator B’s volume slightly to support the sawtooth without overpowering it. Optionally, set Oscillator B one octave higher to emphasize the airy overtones.
2. Filter:
   • Use a Lowpass Filter with a Cutoff frequency around 1-2 kHz to keep the sound warm yet bright.
   • Resonance: Set to low-medium (0.2-0.4) for subtle harmonic emphasis without making it too sharp.
   • Apply a slight Filter Envelope to modulate the Cutoff. Set Attack to ~10-20 ms, Decay to ~500 ms, Sustain at 50-70%, and Release at 100-200 ms.
3. Envelopes:
   • Amplitude Envelope: Short Attack (0-10 ms), medium Decay (~500 ms), Sustain at 70-80%, and short Release (100-200 ms) for a crisp yet smooth sound.
   • Assign the envelope to the Filter Cutoff for dynamic movement.
4. Modulation:
   • LFO: Assign an LFO with a slow Rate (0.5-1 Hz) and Triangle Waveform to modulate the Filter Cutoff or Pan for subtle movement, mimicking the string layer’s texture.
   • Optional: Add slight Pitch modulation (depth ~0.05-0.1) for a livelier sound.
5. Effects:
   • Chorus: Enable Serum’s Chorus with a slow Rate (0.3-0.5 Hz) and medium Depth for a vintage, wide feel.
   • Reverb: Add a light Reverb with a short Decay (1-2 seconds) and low Mix (10-20%) to maintain the disco vibe.
   • Delay: Use a subtle Stereo Delay (1/8 or 1/16 note) with low Feedback (20-30%) for added depth.
6. String Layer:
   • To emulate the Yamaha SS-30 String Machine, create a second Serum patch or use a separate string VST. Choose a soft string waveform (e.g., “Analog Strings” in Serum) with a longer Attack (50-100 ms) for a smooth string texture.
   • Mix this layer quietly beneath the lead to support it without overpowering.

Additional Notes

• Reference: The lead sound was primarily created with an ARP Odyssey, layered with a Yamaha SS-30 String Machine. The bass combines a Minimoog and bass guitar, which you could recreate separately if needed.
• Tutorial and Presets: A YouTube tutorial provides detailed Serum settings and a free preset for this sound: YouTube - How to Make ABBA "Gimme Gimme Gimme" Lead Synth. This is a great starting point.
• MIDI Data: Professional MIDI files for the song’s melody are available at sites like midi.com.au, which you can import into your DAW for accurate note reproduction.

Limitations
I don’t have direct access to a specific Serum preset file (e.g., .fxp) for this sound, as these are often proprietary or created by third parties. However, you can build the patch using the settings above or download the preset from the linked YouTube tutorial. If you need MIDI data or further assistance with specific Serum parameters, let me know, and I can guide you further!

Would you like help with a specific part of the process, like fine-tuning the preset or sourcing MIDI files?

 

I read the title wrong, I thought you meant AI to write patches and keygens and such.