Music Philosophy & Production Approach
I produce electronic music with a focus on DUB, Techno, and Reggae, blending traditional production techniques with AI-assisted workflows. For me, production is about the tension between precision and spontaneity. DUB teaches patience and the power of subtraction. Techno shows how minimal elements create maximum impact. Reggae reminds me that groove and feel always matter more than complexity.
Check out my KingOfDUB YouTube channel for production content and tutorials.
AI-Powered Production with Ableton MCP Extended
Ableton MCP Extended is an MCP server I maintain that lets AI assistants control Ableton Live through natural language. It translates creative intent into precise Live actions via Remote Scripts, so I can focus on ideas rather than mouse clicks.
Key capabilities of my enhanced fork:
- Dual TCP + UDP servers (ports 9877/9878) for reliable operations and ultra-low latency parameter updates, roughly 100-250x faster than TCP alone
- Automated dub techno generation: 2-hour sets with 12 tracks, 48 MIDI clips, 30 sections, and timer-based scene progression
- ElevenLabs voice integration: text-to-speech and voice cloning directly in session
- Max for Live audio export: real-time WAV/AIFF/OGG/FLAC capture during performances
- Music theory helpers: scale-aware note generation and genre-specific chord progressions for DUB, Techno, and Reggae
- XY mouse controller: custom real-time parameter control over UDP for filter sweeps, volume, and pan
I use it daily for everything from building rhythmic patterns ("create a 4-bar techno loop with sidechain on the kick") to live performance ("gradually increase reverb send on the pad over 32 bars"). The repository includes setup scripts, example workflows, and full documentation.
ACE-Step 1.5: Open-Source Music Foundation Model
ACE-Step 1.5, developed by ACE Studio and StepFun, is the most capable open-source music generation model available today. Released under the MIT license, it represents a genuine alternative to closed-source services like Suno and Udio.
Architecture
ACE-Step uses a hybrid architecture that separates planning from generation:
- Qwen3 language model acts as a musical planner, understanding text descriptions, lyrics, and structural intent
- Diffusion Transformer (DiT) handles the actual audio generation, producing high-fidelity output from the planner's representations
This split is deliberate. Language models excel at understanding musical concepts, structure, and style. Diffusion models excel at generating coherent audio. Combining them produces results that neither architecture achieves alone.
Capabilities
The model supports a wide range of music generation tasks:
- Text-to-music: describe a track and generate it
- Cover generation: reinterpret existing songs in different styles
- Repaint: selectively regenerate sections of a generated track
- Track separation: isolate stems from generated audio
- Multi-track generation: produce separate instrument tracks
- Vocal2BGM: extract or generate instrumental backing from vocal input
- LoRA fine-tuning: train on as few as 8 songs in about 1 hour on an RTX 3090
Performance & Benchmarks
Two model variants are available: 2B (standard) and 4B XL. The XL variant scores 47.9 combined on standard benchmarks, beating Suno v5's 46.8. Generation is fast: under 2 seconds per song on an A100. For local use, the turbo variant with INT8 quantization runs in under 4GB VRAM, making it practical on consumer hardware.
The model supports lyrics in 50+ languages and is freely available at acemusic.ai.
Local Installation
uv run acestep
A VST3 plugin is also available for DAW integration. Full source and documentation at github.com/ace-step/ACE-Step-1.5.
ACE-Step fits naturally into my workflow. I use it for rapid prototyping of ideas, generating reference tracks, and exploring sonic directions I wouldn't reach through traditional means. Combined with Ableton MCP Extended, it creates a pipeline from text description to finished arrangement.
Music Theory & Production Techniques
Understanding rhythm, harmony, and genre-specific techniques is essential for authentic production. Each genre has its own DNA.
| Genre | BPM Range | Common Keys | Signature Techniques | Production Essentials | |-------|-----------|-------------|---------------------|----------------------| | DUB | 60-90 BPM | Minor keys, F minor, A minor | One Drop rhythm, heavy bass subversion, echo/delay effects, sparse arrangement, filter sweeps | Spring Reverb, Tape Delay, Analog Mixer, Saturation | | Techno | 120-150 BPM | A minor, D minor, E minor | 4/4 kick patterns, sidechain compression, hypnotic loops, minimal arrangement, build tension through automation | Roland TR-909, TB-303, Analog Synths, Filters | | Reggae | 65-80 BPM | F major, G major, C major | Offbeat rhythm (skank), roots bassline, one-drop drum pattern, organ bubble | Skank guitar, Hammond organ, Bass guitar |
Genres & Style
DUB
My DUB productions draw from the Jamaican tradition of versioning and mixing desk experimentation. Heavy delay, reverb, and filter sweeps create that cavernous, psychedelic sound. DUB is about space and subtraction. I often strip arrangements to their essentials, letting echoes and reverberations fill the void.
Techno
My techno work focuses on hypnotic, driving rhythms with subtle textural evolution. Every element serves a purpose, building tension through careful automation and spatial processing. Small changes over long periods create trance-like states.
Reggae
Rooted in the one-drop tradition, my reggae productions emphasize groove and space. The rhythm should feel natural and relaxed, even when the production is precise. I focus on the interplay between instruments, letting each element support the others.
Resources
Essential Listening:
- DUB: King Tubby, Scientist, Lee "Scratch" Perry, Augustus Pablo
- Techno: Derrick May, Jeff Mills, Juan Atkins, Robert Hood
- Reggae: Bob Marley, Peter Tosh, Burning Spear, Culture
Production Tools I Use:
- Ableton Live, Native Instruments Komplete, Valhalla DSP, FabFilter
Interested in collaboration or discussing AI-assisted music production?