Projects
PhD thesis
- Bittersweet Lessons in Music AI Research KTH Royal Institute of Technology, 2026
Papers and projects
First author
- SMART: Tuning a Symbolic Music Generation System with an Audio Domain Aesthetic Reward Use an audio model of human preferences as a reward to tune a symbolic music generator. AIMC 2025
- Retrieval Augmented Generation of Symbolic Music with LLMs Generate and edit music from text with an LLM that retrieves relevant examples to learn from. ISMIR 2024 Late-Breaking Demo
- Steer-by-Prior Editing of Symbolic Music Loops Superposed language modelling: edit music loops by constraining which values each note attribute may take. MML 2024
- DDSP-based Neural Waveform Synthesis of Polyphonic Guitar Performance from String-wise MIDI Input Synthesize realistic polyphonic guitar audio from per-string MIDI. DAFx 2024
- SYMPLEX: Controllable Symbolic Music Generation using Simplex Diffusion with Vocabulary Priors A diffusion model for symbolic music that you steer by restricting its vocabulary. arXiv 2024
- Exploring Softly Masked Language Modelling for Controllable Symbolic Music Generation Soft masks let a masked language model take partial constraints on the music it generates. arXiv 2023
- Audio Latent Space Cartography Draw maps of a sound collection's latent space using TimbreCLIP. ISMIR 2022 Late-Breaking Demo
- TimbreCLIP: Connecting Timbre to Text and Images A shared embedding of instrument sounds, text and images, used to control an equalizer with text. arXiv 2022
- Neural Music Instrument Cloning From Few Samples Clone an instrument with a neural synthesizer from as little as 16 seconds of audio. DAFx 2022
- The control-synthesis approach for making expressive and controllable neural music synthesizers Turn DDSP models into MIDI-playable synthesizers trained on unannotated audio. AIMC 2020
Side projects
- Birds drawn by code Procedural birds you blend by dragging across a map, with the code that draws them. 2026
- vibejam A browser music environment where your coding agent composes music as JavaScript. 2025
- Soundfont Generator Generate soundfonts with latent flow matching. 2024
- Gold-Caps_LMD-Matched_General GPT-4 captions for about 30,000 tracks in the Lakh MIDI dataset. Dataset, 2023
- TextSynth Generate synthesizer sounds from text prompts with a simple evolutionary algorithm. 2023
- Loop Savant A loop-based music making tool that runs in your browser. 2023
Contributing author
- Exploring the Expressive Space of an Articulatory Vocal Model using Quality-Diversity Optimization with Multimodal Embeddings Search the sounds of a vocal tract model, steered by text prompts. GECCO 2025 · Joris Grouwels, Nicolas Jonason, Bob L. T. Sturm
- Sparks of Musical AGI? Challenges and perspectives in music co-creation with LLMs A qualitative look at what LLMs know about music and how well they help make it. AIMC 2024 · Luca Casini, Nicolas Jonason, Bob L. T. Sturm
- Applying textual inversion to control and personalize text-to-music models Teach MusicGen new words from a few short audio examples. MML 2024 · Carl Thomé, Bob L. T. Sturm, John Pertoft, Nicolas Jonason
- Investigating the viability of masked language modeling for symbolic music generation in abc-notation Masked language models for Irish traditional music in abc notation. EvoMUSART 2024 · Luca Casini, Nicolas Jonason, Bob L. T. Sturm
- Generating folk-like music in abc-notation with Masked Language Models abcMLM: a masked language model that generates folk tunes in any order. ISMIR 2023 Late-Breaking Demo · Luca Casini, Nicolas Jonason, Bob L. T. Sturm