Category: Planning

Track Inspiration

I’m a child of 90’s dance music. A lot of what I like is based on that period of music. Hence my first track for the EP (currently named “Latent Space”) is a homage to the Robert Miles track “Children” (1995).

The track didn’t start out that way, but it morphed into something like it over a period of several weeks.

The first version used a simple synth. It was just a sketch, so it was fine that it was using almost anything. There was almost no arrangement (Figure 1). The drums were just a loop, and the bass and pads were simple.

[LINK TO “Track1-v1(edit).mp3”]

https://artslondon-my.sharepoint.com/:u:/g/personal/t_shepherd0320221_arts_ac_uk/EedF1rSD4OhKnlRz2Q29nXcBW60h4_Uoc0Dm7CGYUnBUTA?e=2aKBlq

**Figure 1:** Track 1 | Version 1 | Almost no arrangement, simple instrumentation, not much happening.

By version two of the track, you can see a more defined arrangement (Figure 2). The main lead sound now uses Ableton “Tension” plugin for a piano-like sound (Figure 3). The piano is a little distorted (though the use of the “Pedal” plugin. And while not 100% right, it was closer than the original sound.

[LINK TO “Track1-v2”]

https://artslondon-my.sharepoint.com/:u:/g/personal/t_shepherd0320221_arts_ac_uk/Ee_z–KvIEVCibUqVlq30TUBd3epRY21dPFtIRYPzrviKw?e=MrJwf3

**Figure 2:** More arrangement, first use of a piano-like sound, and better bass.

**Figure 3:** Use of “Tension” Plugin from Ableton 12 Suite for the “Piano” sound.

From version two to version seven I still used the distorted “piano” sound. With feedback from several people, I swapped the distorted piano for a sample-based one, in this instance “The Grandeur” from Native Instruments (Figure 4).

[LINK TO Track1-v8]

https://artslondon-my.sharepoint.com/:u:/g/personal/t_shepherd0320221_arts_ac_uk/EXY60HwfvV1GhpOoDWHb1REBpA-T_phQLxdSQbT5mfKd7A?e=PmjuDa

**Figure 4:** “The Grandeur” from Native Instruments.

This piano was coupled with the Raum reverb plugin to put the piano into a larger space. It was at this time that I realised that I was creating a homage to “Children”. For the most part this was a subconscious decision. But it just works. So why does it work?

Looking at the frequency range of instruments

When considering sound design in the process of generating tracks it is sensible to keep in mind their frequency distribution. This can help in choosing instruments that don’t fight to be heard in a mix. Thankfully there are plenty of freely available charts on the internet to help with this. Figure 5 shows one from Sweetwater that will help me to illustrate why this track works, and why Children (1995) from Robert Miles worked.

Figure 5: Musical Instrument Frequency Cheat Sheet (Author: Sweetwater, 2024)

While a good start, I’ve pulled out the overlapping frequencies of the instruments (Figure 6) to better illustrate how the instruments overlap.

**Figure 6:** How the different instruments frequency ranges overlap in key areas.

The main thing to realise here is that as there is no vocal in this track, the piano isn’t masking a vocal. The main issues masking of the snare by the piano. And a lot of overcrowding in the high mid-range and the frequencies over 5K.

These issues are easier to fix without having to worry about a masked vocal. For the snare, I reduced around 3-5dB at around 300k. This preserves the body of the snare. In the mid-range I reduced the snare to remove some of the “boxiness” and make way for the piano to come to the front of the mix. In the highs it was a balance to keep the presence of the piano versus the snare and the hi-hats.

In the end, while the frequency ranges can help, I ultimately did this all by ear and trusted what I was hearing in my studio monitors. Though I do like how the chart helps to contextualise the decisions that I made with my ears.

Bibliography

Miles, R. (1995) “Children”. Available at: Apple Music (Accessed: 11 January 2025)

Sweetwater (2024) EQ Frequencies of Musical Instruments Explained. Available at: https://www.sweetwater.com/insync/music-instrument-frequency-cheatsheet/ (Accessed: 20 February 2025)

Planning

How am I feeling?

I’ve found the process for this EP has run across the entire spectrum of emotions. From dread to happiness, and everything in-between.

At the start of the process, there was the initial fear that I’d taken on a project that may fail. This was due to hearing the initial output of the AI generated midi. It sounded clunky, and I wasn’t sure that I’d be able to get anything sounding good from it. However, there were a few reasons for the clunky sounds. One, the velocity and the timing were all over the place. In the end I learnt that I was fighting against the output. And two, I couldn’t work out if the output was an issue with my training data, or the way I made the model, or both. It was very stressful, and I didn’t want to be stuck doing development work when I should be focusing on music production.

The other issue, and it was a big one, was the amount of data I needed to create to get this working. I thought I had a handle on it, it was going to be a lot. What I didn’t realise was just how much (Figure 1).

**Figure 1:** A screenshot of just some of the midi files used to create the ST4RT+ dataset

To get these all working I created a master Logic file (Figure 2) with all the midi files in there. Then had to export them one at a time. It was a tedious process to get it all completed.

**Figure 2:** Master Logic Pro X session with all of the midi ready to export.

I wanted the midi output to be more structured and considered. But try as I may, it just was not working. After I embraced the nature of what the AI was producing then I started to think more like a music producer and less like the AI would solve issues for me, this way of working, led me to the first track that I liked.

To me, this highlights what a music producer is and does. You work with artists to capture, edit, manipulate, arrange, and preserve (Slade, et.al., 2020) their ideas and make them better. The keywords here are: manipulate and arrange. By taking the generated melody and modifying the chunks into phrases, I found that I could get something – not only great – but amazing.

From taking raw material and working with an artist to create something that captures the feeling that the artists is trying to achieve.

I think that the main reason that I was having issues initially not seeing my role in that way in the early part of the process. Once that changed, I was easier able to create work that feels right and works with the methodology created for this project.

Bibliography

Slade, S., Thompson, D. Massy, S., Webber, S. (2020) Music Production: What Does a Music Producer Do? Available at: https://online.berklee.edu/takenote/music-production-what-does-a-music-producer-do/ (Accessed: 27 February, 2025)

Creative Planning

Sound Advice

I’ve been reflecting on the process of creating these tracks over the past few weeks, and one of the main things that I’ve realised is the importance of sound/instrument selection.

It has also been one of the hardest things to get right. For instance, Listen to the following (the main instrument for the melody at the start):

[LINK TO Track3.wav]

https://artslondon-my.sharepoint.com/:u:/g/personal/t_shepherd0320221_arts_ac_uk/EVqdlEVAY75MlRjiLQwzT7oB8w8hWXa17gctKCvDDWW90Q?e=c2bInb

Now, listen to the same segment with different instrument selection:

[LINK TO Track3-v11.wav]

https://artslondon-my.sharepoint.com/:u:/g/personal/t_shepherd0320221_arts_ac_uk/EdX5G-xOjQBOoIXS-tFIfW0BYmVFCRTCFJ9WKrn0ABLgFw?e=2DTe57

The original lead sound doesn’t have as much presence in the mix. So, why does the second one sound better?

I could have introduced EQ, compression, or saturation, to get the sound to sit better in the mix, and I will say, I tried that first. The results were fine. But I really wanted the lead sound to cut through. So, I decided to look for similar sounds that would sit better in the mix. It takes ages to do this, but it was a good way to keep myself overworking a sound.

It’s sometimes better to change the sound rather than trying to make incremental changes to get it to sound right. So, for that lead sound I changed the sound from (Figure 1):

**Figure 1:** Lead sound used in “Track3-v1”

To (Figure 2):

**Figure 2:** Lead sound used in “Track3-v11”

The original sound is a “purer” synth sound. And I loved the pitch shift at the start of the notes, but it didn’t cut through the mix enough. The second sound has more distortion but also a more organic feel.

Other techniques used for sounds

Layering

I’ve also used a few techniques to get the best from a few sounds where using the stock plugins could achieve a sound that I was going for by itself. For these types of sounds I would sometimes layer them together (Figure 3).

**Figure 3:** See the three tracks have the same notes with different instruments all playing at the same time.

Using layering, I was able to create and treat the attack of an instrument that I liked with the sustain of a different instrument that I liked too. While in Figure 3 the instruments are on separate tracks, I could have used an instrument rack. I have done this for a few of the tracks as it made automation easier, but for the example above it wasn’t necessary to achieve the sound I was after. And in this instance, I was using different octaves for the different layers, so keeping them separated made turning them into one instrument easier.

Modulation

I can’t say enough about how modulation can help with creating aural interest to sounds over time. While it’s possible to do this with automation of parameters using an automation lane, it’s usually a lot faster than easier to use an LFO.

I can say though, there are a few drawbacks with using the LFO in Ableton. First, if it’s not synced to the grid (using frequency rather than note values), you can end up playing a version that is amazing but when you render it out, the start value of the frequency is in a different place every time you render it out. Meaning it never sounds the same twice. This doesn’t happen when the LFO is set to the grid.

But an easy way around this is to freeze the track, see if it’s good, if not unfreeze and refreeze, listen and repeat until you are happy. Not ideal, but better than it sounding different every time.

Creative Development Planning Research

Can you collaborate with an ANT?

Post author By Tim Shepherd
Post date 28th February 2025
No Comments on Can you collaborate with an ANT?

Just kidding. But it does bring up the question of how do I collaborate with a non-human entity? Is it even possible?

Well, from my perspective it is. However, let’s look at what others in the fields of creativity and sociology say.

I’ll start with a sociological perspective. A definition of Actor Network Theory (ANT) is a good place to start so that we can break down how this relates to my Final Major Project (FMP) and what insights working with a non-human collaborator bring up.

Actor Network Theory: Is a theoretical framework within the field of science and technology studies (STS) that examines how social, technical, and material entities (referred to as “actors”) interact to form complex networks that shape and influence outcomes. ANT challenges traditional distinctions between human and non-human agents by treating them symmetrically as participants in these networks.

This definition, while not specifically saying that non-human entities can be collaborators as an overt term, states that the interactions between human and non-human “actors” to influence one another.

ANT is a contrasting view to Technological Determinism (TD), where the idea that technology develops independently of social change and drives social change (Bimber, 1990). For example, Karl Marx believed that the railway in colonial India changed social hierarchies by introducing new economic activities (ibid.). While TD can look like a good place to start when you take a cursory view of any technology and its impact on how people use it, I believe a more nuanced approach can lead to a better understanding of how we as humans interact with technology, and how we as humans shape technology. To gain a more holistic view of collaboration I’ll bring up Fraser’s “Collaboration bites back” (2022). In this paper Fraser creates a manifesto for collaboration as a tool for change. So, I thought it best to go through her 10-point manifesto and see/explain how working with ST4RT+ achieves her points.

Collaboration should not be predictable:
This is an easy one. While ST4RT+ is based on my melodies and data, it doesn’t create melodies that are 100% what I would do.
Collaboration should not be clean:
This one is a little more nuanced. I will say that when I was struggling with the outputs of the model at the start of this project, I had to get my hands dirty and get to the point where I started thinking more like a music producer and less like a developer.
Collaboration should not be safe:
This whole project was a risk, using technology I’d never used before, and risking that it was going to work has put me in a place where I thought I was going to be lucky to generate anything worthwhile.
Collaboration requires consent:
Harder to do this with a non-human collaborator, however if the original generation of a set of melodies is objectively awful (all the notes are overlapped and on bar one) then I just regenerate.
Collaboration requires trust:
This point is interesting, for me it was about trusting myself and the process. When I was fighting the models output it was because I wasn’t trusting my skills as a music producer. I wanted the model to generate clean melody lines. Trust in myself has really helped to get this project working.
Collaboration requires time, and time (usually) costs money:
This project has taken time to get working (far more time in the beginning than I anticipated). It has needed experimentation and failure to get to a point where the process and methodology are working.
Collaboration requires vigilance:
Regardless of a non-human collaborator, this still applies, though it relies more on me to do that work.
Collaboration is not compulsory:
Nothing to see here… in this case it was compulsory.
Collaboration is not cool:
I disagree here. Only because using an ANT framework almost everything is a collaboration even if you aren’t aware of it.
Collaboration is a tool for change:
I agree that any collaboration should challenge the status quo. For me the idea of creating an ethical use for AI trained only on the data that I have given it challenges how AI is being used and the data it is trained on. For me this is important and a point of difference with this project.

I think that when I look at Fraser’s 10-point manifesto that this project still works in terms of meeting what she defines as collaboration.

Bibliography

Bimber, B. (1990) Karl Marx and the Three Faces of Technological Determinism, in Social Studies of Science, Vol. 20, No. 2 (May, 1990), pp. 333-352. Available at: https://www.jstor.org/stable/285094 (Accessed: 2 December 2024).

Fraser, J. (2022) Collaboration bites back. Available at: https://www.julietfraser.co.uk/app/download/11414030/Collaboration+bites+back.pdf (Accessed: 18 October 2024)

Planning

Methodology

The journey for this project so far has been experimental to say the least. Fortunately, the methodology for creating and producing the work has been less experimental.

The practice-as-research methodology employed in the production of the artefacts engages several phases of development:

Composition of musical material (Melody and Harmony) as input.
Generation of the output with the (AI) collaborator and reflection on it.
Development of the artefacts with a DAW.

While there is nothing unusual about the development phases compared to a more run-of-the-mill collaboration, the composition relies on a non-human participant, evoking Actor-Network Theory (ANT). This runs counter to Technological Determinism, which would posit that the technology itself is having an impact on the output, rather than the complex interactions of the inputs to the model (a technology), the model (another technology, the output, my ability to have some control over the output via weights, and any other human and non-human factors that create a complex network of relationships that can affect the final product.

To make this a lot easier to digest, I’ve created a 5-step methodology (Figure 1) that captures the process of working with my AI collaborative partner though to producing a track.

**Figure 1:** 5-Step Methodology for creating tracks for the ST4RT+ EP.

The breakdown of the phases goes as follow:

Generate a melody with ST4RT+ based on the input from my written melody and harmony.
Check the output from the model, if the output sounds wrong, modify the output weights and regenerate if necessary.
Document the weights used to create the output melody. These weights may be different for each melody to create a good output.
Generate a second melody (melody only) that is based on the first input using the same weights as last time.
Develop the materials in a DAW, reflect on the output and regenerate if necessary.

The regeneration parts of this methodology aren’t deleterious, in this case meaning that I keep all the previous outputs as both reference and as opportunities for remixing with other outputs. I’d do this with a human collaborator, so I may as well do it with the AI model. I guess the best way to look at this is if a human collaborator came up with an amazing guitar riff for a song, but the verses were terrible, you wouldn’t throw out the amazing riff because the verse was bad. You’d simply work on getting the verses better.

In this way, the AI has felt like a collaboration partner. Sometimes it gets it right. Other times I feel it misses the mark, but by working with it (rather than against it) it can sometimes surprise and delight when we find a solution that is better than the sum of its parts.

Recent Posts

Categories