Searcharxiv⌕ Search

arXiv subjects

Wanying Wu

Publications and source records attributed to Wanying Wu.

3 recordsLinked to original sources

Decomposing the Associative Chain: Schema-Based Mediation Supports Expression and Understanding in Human-AI Music Co-creation

Users cannot reliably translate their creative intents into AI-generated music. We contend that the root of this difficulty lies in associative divergence: associations between narratives and musical features are overly remote, and users lack a visible mediation structure to negotiate them. To address this, we introduce five core schema groups as cognitive mediation and propose a bidirectional schema-based framework. Its forward pathway decomposes narrative-to-music associations into staged mappings, while its reverse pathway provides structured interpretations from music back to narrative. Findings from two user studies(N = 188; N = 60) show that the framework led to clearer creative expression and greater perceived control, improved the comprehensibility of generated music, especially for unconventional pieces. Furthermore, it supported cultural and affective thinking while helping users locate associative divergence. Our work points toward better creative intent alignment in human-AI music co-creation.

cs.HC↗

From Facts to Insights: A Persona-Driven Dual Memory Framework and Dataset for Role-Playing Agents

While role-playing agents excel in short-term interactions, long-term conversations overwhelm context windows, motivating external memory frameworks. Current systems typically rely on persona-agnostic summarization, which records facts without persona-specific interpretation, yielding generic responses that compromise persona fidelity. To bridge this gap, we introduce RoleMemo, a dataset featuring four reasoning tasks where the factual fragments must be interpreted through the persona to reach the correct answer. Evaluation on RoleMemo exposes critical limitations of persona-agnostic frameworks. We thus propose DualMem, which decouples memory into two streams: factual cognition and persona-conditioned insight. Trained through Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL), our framework with a 4B-parameter model outperforms zero-shot persona-agnostic frameworks powered by DeepSeek-V3.2 for sustained persona fidelity. Our resources are available at https://github.com/Collab-Gen/RoleMemo.

cs.CL↗

MusicLayout: Explicit Structural Planning for Controllable Text-to-Music Generation

Text-to-music generation has advanced rapidly, but current systems still rely primarily on global text prompts, leaving the structural organization of generated music implicit and difficult to inspect, control, or revise before audio generation. To address this issue, we introduce MusicLayout, an explicit intermediate representation for controlling musical structure in text-to-music generation. MusicLayout describes a musical piece as a time-aligned layout of sections, textures, repetitions, variations, and instrument-level arrangements, serving as an interpretable planning layer between textual intent and the generated music. We integrate MusicLayout into a text-to-music framework built on a unified autoregressive formulation, where the model first generates a MusicLayout representation and subsequently predicts audio tokens conditioned on this representation within a single sequence. The resulting MusicLayout can be inspected and modified prior to audio generation, providing a mechanism for layout-level structural control. We evaluate MusicLayout through layout-conditioned generation, layout manipulation experiments, and matched-data ablations, providing evidence that explicit layout planning can improve long-range structural organization and support layout-level control. We have released the implementation as open source on GitHub at https://github.com/XaryLee/MusicLayout.

cs.SD↗