{"slug": "gala-generation-aware-cross-modal-alignment-for-text-to-time-series-synthesis", "title": "GALA: Generation-Aware Cross-Modal Alignment for Text-to-Time-Series Synthesis", "summary": "Researchers introduced GALA, a two-stage generation-aware cross-modal alignment method for text-to-time-series synthesis, which sets a new state of the art on the TSFragment-600K benchmark, ranking first in 30 of 36 metric columns with average ranks of 1.08/1.08/1.42 at lengths 24/48/96, outperforming the strongest baseline's 1.92/2.00/1.75. The method couples a pretrained text encoder with a time-series foundation model into a shared embedding space, adapted by an auxiliary generative loss, and freezes the resulting caption embedding to drive a flow-matching generator, breaking the trade-off between fidelity and caption adherence.", "body_md": "arXiv:2608.13741v1 Announce Type: new\nAbstract: Synthesizing time series from natural language is emerging as the most expressive form of controllable time series generation. However, existing text-conditioned generators either take caption embeddings frozen from off-the-shelf text encoders, or adapt the encoder end-to-end, letting the denoising loss shape the embeddings only as a by-product. In either case, the conditioning representation is never deliberately matched to the signal modality, leaving it ill-suited to guide generation. We address this by introducing GALA: Generation-Aware cross-modaL Alignment for text conditional time series generation. GALA is a two-stage approach that first contrastively couples a pretrained text encoder with a time-series foundation model into a shared embedding space with both encoders adapted to generation by an auxiliary generative loss, and then freezes the resulting caption embedding to drive a flow-matching generator. On TSFragment-600K, spanning four domains and three fragment lengths, GALA sets a new state of the art, ranking first in 30 of 36 metric columns and reaching an average rank of 1.08/1.08/1.42 at lengths 24/48/96 against 1.92/2.00/1.75 for the strongest baseline. We further find that generator-internal text encoders force a trade-off between fidelity and caption adherence, whereas conditioning on the aligned embedding breaks it: FID, CTTP, and JFTSD all improve at once. Ablating the auxiliary loss degrades FID, CTTP and JFTSD together, it indicates the generative term is a necessary component of the alignment rather than an add-on.", "url": "https://wpnews.pro/news/gala-generation-aware-cross-modal-alignment-for-text-to-time-series-synthesis", "canonical_source": "https://arxiv.org/abs/2608.13741", "published_at": "2026-08-17 04:00:00+00:00", "updated_at": "2026-08-17 04:14:16.345977+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "generative-ai", "natural-language-processing"], "entities": ["GALA", "TSFragment-600K"], "alternates": {"html": "https://wpnews.pro/news/gala-generation-aware-cross-modal-alignment-for-text-to-time-series-synthesis", "markdown": "https://wpnews.pro/news/gala-generation-aware-cross-modal-alignment-for-text-to-time-series-synthesis.md", "text": "https://wpnews.pro/news/gala-generation-aware-cross-modal-alignment-for-text-to-time-series-synthesis.txt", "jsonld": "https://wpnews.pro/news/gala-generation-aware-cross-modal-alignment-for-text-to-time-series-synthesis.jsonld"}}