@inproceedings{huang-etal-2026-gimflow,
title = "{GIMF}low: An {LLM}-Guided Dialogue-to-Music System for {GIM}-Inspired Well-Being Experiences",
author = "Huang, Ruijia and
Yao, Yao and
Zhao, Hai",
editor = "Agrawal, Sweta and
Abad, Alberto",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-demos.65/",
pages = "685--695",
ISBN = "979-8-89176-461-3",
abstract = "Guided Imagery and Music (GIM) organizes verbal preparation, guided listening, imagery exploration, and reflection into a structured, music-centered process. Existing LLM-based well-being agents remain text-centered, and emotion-aware music systems operate outside such multi-phase workflows, conditioning music on static mood descriptions. We present GIMFlow, an interactive, non-clinical GIM-inspired system that translates reflective dialogue into a structured musical trajectory through four phases: Prelude, Induction, Music Imaging, and Postlude. GIMFlow estimates current and target affective states in continuous valence-arousal space, constructs a deterministic four-waypoint transition, and maps waypoints to sections of an LLM-generated symbolic music blueprint for deterministic rendering; the system also supports metadata-based retrieval and unified session-level logging. A formative within-subject study (N=23) of the initial prototype revealed that generated music captured only the endpoint state; we then aligned waypoints with intra-piece sections and conducted a preliminary within-subject evaluation of the refined system (N=10). Results show descriptively favorable trends in valence and negative affect while positive affect and arousal are mixed; all findings are exploratory. GIMFlow is released as an open-source, installable research demonstration with example sessions and reproducible logging utilities."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="huang-etal-2026-gimflow">
<titleInfo>
<title>GIMFlow: An LLM-Guided Dialogue-to-Music System for GIM-Inspired Well-Being Experiences</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ruijia</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yao</namePart>
<namePart type="family">Yao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hai</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: System Demonstrations</title>
</titleInfo>
<name type="personal">
<namePart type="given">Sweta</namePart>
<namePart type="family">Agrawal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alberto</namePart>
<namePart type="family">Abad</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-461-3</identifier>
</relatedItem>
<abstract>Guided Imagery and Music (GIM) organizes verbal preparation, guided listening, imagery exploration, and reflection into a structured, music-centered process. Existing LLM-based well-being agents remain text-centered, and emotion-aware music systems operate outside such multi-phase workflows, conditioning music on static mood descriptions. We present GIMFlow, an interactive, non-clinical GIM-inspired system that translates reflective dialogue into a structured musical trajectory through four phases: Prelude, Induction, Music Imaging, and Postlude. GIMFlow estimates current and target affective states in continuous valence-arousal space, constructs a deterministic four-waypoint transition, and maps waypoints to sections of an LLM-generated symbolic music blueprint for deterministic rendering; the system also supports metadata-based retrieval and unified session-level logging. A formative within-subject study (N=23) of the initial prototype revealed that generated music captured only the endpoint state; we then aligned waypoints with intra-piece sections and conducted a preliminary within-subject evaluation of the refined system (N=10). Results show descriptively favorable trends in valence and negative affect while positive affect and arousal are mixed; all findings are exploratory. GIMFlow is released as an open-source, installable research demonstration with example sessions and reproducible logging utilities.</abstract>
<identifier type="citekey">huang-etal-2026-gimflow</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-demos.65/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>685</start>
<end>695</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T GIMFlow: An LLM-Guided Dialogue-to-Music System for GIM-Inspired Well-Being Experiences
%A Huang, Ruijia
%A Yao, Yao
%A Zhao, Hai
%Y Agrawal, Sweta
%Y Abad, Alberto
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: System Demonstrations
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-461-3
%F huang-etal-2026-gimflow
%X Guided Imagery and Music (GIM) organizes verbal preparation, guided listening, imagery exploration, and reflection into a structured, music-centered process. Existing LLM-based well-being agents remain text-centered, and emotion-aware music systems operate outside such multi-phase workflows, conditioning music on static mood descriptions. We present GIMFlow, an interactive, non-clinical GIM-inspired system that translates reflective dialogue into a structured musical trajectory through four phases: Prelude, Induction, Music Imaging, and Postlude. GIMFlow estimates current and target affective states in continuous valence-arousal space, constructs a deterministic four-waypoint transition, and maps waypoints to sections of an LLM-generated symbolic music blueprint for deterministic rendering; the system also supports metadata-based retrieval and unified session-level logging. A formative within-subject study (N=23) of the initial prototype revealed that generated music captured only the endpoint state; we then aligned waypoints with intra-piece sections and conducted a preliminary within-subject evaluation of the refined system (N=10). Results show descriptively favorable trends in valence and negative affect while positive affect and arousal are mixed; all findings are exploratory. GIMFlow is released as an open-source, installable research demonstration with example sessions and reproducible logging utilities.
%U https://aclanthology.org/2026.emnlp-demos.65/
%P 685-695
Markdown (Informal)
[GIMFlow: An LLM-Guided Dialogue-to-Music System for GIM-Inspired Well-Being Experiences](https://aclanthology.org/2026.emnlp-demos.65/) (Huang et al., EMNLP 2026)
ACL