Large Language Models (LLMs) have demonstrated remarkable proficiency in diverse tasks. This success raises a fundamental question in machine composition: Can symbolic music be considered a special form of language that can be jointly modeled with natural language for composition tasks? Recent studies validate that symbolic music can be modeled as a human language, yet composing structured music from partial symbolic inputs through natural language interaction remains underexplored. Even LLMs struggle to generate structurally coherent compositions in such hybrid input-output scenarios, highlighting a fundamental gap that calls for a domain-specific learning paradigm. To this end, we propose Inspiration-to-Structure (IoS), a cognitively inspired framework that enables LLMs to generate structured musical sections from melodic ideas. IoS employs a three-phase process—semantic, structural, and collaborative cognition—and is supported by two key components: (1) a new dataset and construction protocol called Structured Triplet Data (STD), and (2) a training method, Dual-Instance Structural Contrastive Optimization (DiSCO), designed to enhance structural awareness. Experiments show that IoS improves structural coherence by 47.8% and artistic creativity by 21.8% compared to conventional language modeling paradigm, supervised fine-tuning, and even enables smaller LLMs to surpass larger LLMs. These results suggest that symbolic music, while language-like, demands specialized modeling beyond standard language modeling paradigms. IoS enables LLMs to transform music theory knowledge into structured composition, empowering users to compose music interactively via language and advancing toward general creative AI.

Abstract The study focused on the application value of the space compounding algorithm‐based colour Doppler ultrasound twinkling artefacts in predicting chemical components of urinary tract stones. Specifically, the minimum variance method was used to optimize the space‐compounding algorithm to improve the quality of colour Doppler ultrasound artefacts. Then, the urinary system stone model was built to quantitatively analyse the influence of stone components on colour Doppler twinkling artefacts. The results showed that no notable differences were noted in the twinkling artefact intensity (TAI), the width of twinkling artefact (TAW), and the length of twinkling artefact (TAL) between the oxalic acid monohydrate, hydroxyapatite, uric acid, and cystine stone models (p < 0.01). The area under the curve (AUC) of the TAI for the diagnosis of calcium‐free calculus models was 0.978, that of TAW was 0.9462, and that of TAL was 0.9315. The diagnostic sensitivity and specificity of TAI were 89% and 96%, respectively, the diagnostic sensitivity and specificity of TAW were 86% and 80%, respectively, and the diagnostic sensitivity and specificity of TAL were 78% and 97%, respectively. The ability of TAL (AUC of 0.953) to distinguish between hydroxyapatite stone model and cystine stone was better than TAI (AUC of 0.702) and TAW (AUC of 0.657) (p < 0.01); and TAL and TAW were positively correlated with TAI (R2 was 0.79 and 0.68, respectively, p < 0.01). In conclusion, the space compounding algorithm based on minimum variance method can improve the quality of colour Doppler ultrasound twinkling artefacts, and colour Doppler ultrasound twinkling artefacts can predict the chemical components of urinary stones, which can be used as a criterion for clinical diagnosis.

Anxiety is the most common mental problem that affects nearly 300 million individuals worldwide. The situation is even worse recently. In clinical practice, music therapy has been used for more than forty years because of its effectiveness and few side effects in emotion regulation. This paper proposes a novel style transfer model to generate the therapeutic music according to user's preference. It is widely recognized that the favorite music greatly increases the engagement of the user, hence results in much better curative effects. But in general, users can provide only one or several favorite songs, which are insufficient for the customization of therapeutic music. To address this difficulty, a new domain adaption algorithm that transfers the learning result for music genre classification to the music personalization, is designed. Targeting the joint minimization of the loss functions, three convolutional neural networks are utilized to generate the therapeutic music with only one labelled data of favorite song. The experiment on the anxiety suffers shows that the customized therapeutic music has achieved better and stable performance in anxiety reduction.