The DOI matches its Crossref record; 48 of the 48 references it cites were found in a registry or an index; 3 of the 15 sources its claims cite were checked for support.
AI-assisted literary-to-song adaptation: a practice-based study of literary interpretation in a text-to-audio workflow
Open accessAbstract
This qualitative practice-based digital humanities article examines AI-assisted literary-to-song adaptation. GenAI research has addressed writing support, music generation, literary analysis and creative work, with less attention to literary judgement in generated audio and post-production. The corpus comprises 32 songs developed by a single creator, largely based on H. P. Lovecraft and produced in a post-punk/indie-rock idiom through close reading, researcher-written lyrics, ChatGPT-supported revision, Udio text-to-audio generation and human-directed assembly. Two cases are analysed: ‘The Viol Screams’ adapts The Music of Erich Zann and ‘Whispers from Yuggoth’ brings together selected sonnets from Fungi from Yuggoth. Benchmarking, listener testing and human-only comparison fall outside scope. The analysis traces source selection, lyric development, generation, evaluation, repair and post-production, focusing on textual transformation, interpretive emphasis and production intervention. Across the cases, generated audio was selected and assembled using literary, musical and technical criteria, while source structure affected human work across stages. Researchers in digital humanities and human-AI creative practice can use the source-to-song workflow heuristic and lyric-strategy typology to document how literary meaning moves into generated sound. Distributed interpretive control gives adaptation and popular music scholars a case-derived concept for examining how interpretive priorities remain active in lyric, vocal, sectional and production decisions.
Overview
CoChat Summary
TL;DR
This practice-based digital humanities article documents a single creator’s AI-assisted workflow for adapting public-domain literary texts into 32 songs, using close reading, researcher-written lyrics, selective ChatGPT line revision, Udio text-to-audio generation, and human-directed DAW assembly. Through two focal cases—‘The Viol Screams’ from *The Music of Erich Zann* and ‘Whispers from Yuggoth’ from *Fungi from Yuggoth*—it argues that literary interpretation remains distributed across source selection, lyric adaptation, generated-audio evaluation/repair, and post-production.
Key findings
- The corpus comprises 32 AI-assisted songs created April 2024–May 2026; 28 adapt H. P. Lovecraft, and all were produced in a post-punk/indie-rock idiom. Completed songs ranged from 3:46 to 6:25, averaging approximately 5:05.
- Five recurrent workflow stages were reconstructed: source-text selection, lyric adaptation, generative music production, iterative selection and repair, and post-production assembly. Five primary lyric-adaptation strategies were identified across the corpus: selective excerpting (10 songs), near-verbatim adaptation (7), expansion (7), partial rewriting (5), and cross-text synthesis (3).
- In ‘The Viol Screams’, the prose story’s delayed window revelation was relocated into a repeated chorus (1:36–2:01, 2:50–3:14, 4:28–4:54) and the song ends with an isolated “Death!” at 5:18, increasing performability while reducing gradual uncertainty and narrative hesitation.
- In ‘Whispers from Yuggoth’, selected sonnets I, II, III, IV, V, XXII, and XXI were ordered into a progression from discovery and pursuit to self-loss, cosmic encounter, and destruction; the chorus recurs at 1:07–1:32, 2:08–2:34, and 4:18–4:44, and the final repeated “die” at 5:28–5:52 closes the track. The adaptation gains coherence but narrows the source sequence’s fragmentary openness.
- The article introduces “distributed interpretive control” to describe how source-derived interpretive priorities remain active across close reading, lyric design, candidate evaluation, audio repair, and post-production.
Methods
- Practice-based qualitative single-creator study combining a corpus-level description of 32 completed songs with retrospective case studies of two purposively selected focal adaptations: ‘The Viol Screams’ (prose compression) and ‘Whispers from Yuggoth’ (cross-text poetic synthesis).
- Production workflow used close reading, researcher-written lyrics, selective line-level ChatGPT revision, Udio text-to-audio generation in 32-second clips with Context Length set to 130 seconds, Prompt Strength 50%, Lyrics Strength 50%, and Clarity 25%, followed by Adobe Audition editing/assembly; LALAL.AI stem separation was used on 16 songs. The corpus was produced across Udio v.1.0 (14 songs), v.1.5 (8 songs), and v.1.5 Allegro (10 songs).
- Analysis reconstructed a five-stage workflow, assigned one primary lyric strategy per song, and mapped adaptation episodes using three dimensions: textual transformation, interpretive emphasis, and production intervention. Benchmarking, listener testing, and human-only comparison fell outside scope.
Limitations
- Stated: the researcher was also the creator, creating risk of confirmation bias; the study included no independent coding, second-scholar review, or music-producer assessment.
- Stated: the production archive is incomplete, especially for early exploratory generations; deleted candidates and complete rejection histories cannot be reconstructed, so reliable generation counts, rejection/repair rates, and precise time-on-task measures are not supported. Complete ChatGPT prompt logs, interaction counts, and suggestion acceptance rates were not retained.
- Stated: the corpus is purposive and stylistically bounded—mostly Lovecraft and all post-punk/indie-rock—and the workflow depended on changing Udio releases and features. The study did not test listener interpretation, fidelity judgements, audience response, human-only composition, single-pass generation, alternative TTA platforms, or unsuccessful adaptations in equivalent detail. Evident: findings are case-derived and not generalizable beyond this creator, corpus, idiom, and platform.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
3 Valid · 8 Not valid · 12 Not checked · 227 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Access to generation does not remove the need for human judgement, since rendered outputs still have to be evaluated, revised and organised within a larger creative structure, particularly when fully rendered audio forms part of an iterative production process (Kim et al., 2025).”
What the sources say
- Partly supportedcites Kim et al., 2025 sourceExperimental — accuracy not yet measured
Section 3.2.3 (p. 4) says participants valued manual songwriting and the ability to choose where to use and edit AI suggestions, supporting the need for human judgement. But the paper discusses editing modular musical elements, not evaluating and revising fully rendered audio within an iterative production process; the Introduction (p. 1) says completed audio songs are less supportive of that kind of iteration. (LLM, conf=0.86)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Across the cases, generated audio was selected and assembled using literary, musical and technical criteria, while source structure affected human work across stages.”
plainn = 2- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“They represent distinct adaptation problems within the wider corpus.”
plainn = 2- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“A musically appealing candidate could still be rejected when the vocal rendering changed words, obscured a key phrase, or weakened the intended relation to the source.”
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The generative process was iterative.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
51 more claims
Create a free account to see all 250 claims and every source verdict.
- It may state two separate claims as one. 80
- Probably not a substantial claim. 174
- The quote only partly states this claim. 2
Citations
Cites
Qualitative Case Study Methodology: Study Design and Implementation for Novice Researchers
Baxter, P., & Jack, S. (2015) · The Qualitative Report · doi:10.46743/2160-3715/2008.1573 · Cited by 7,594
ChatGPT for good? On opportunities and challenges of large language models for education
Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., et al. (2023) · Learning and Individual Differences · doi:10.1016/j.lindif.2023.102274 · Cited by 7,326
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021) · Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3442188.3445922 · Cited by 6,988
Chatting and cheating: Ensuring academic integrity in the era of ChatGPT
Cotton, D. R. E., Cotton, P. A., & Shipway, J. R. (2024) · Innovations in Education and Teaching International · doi:10.1080/14703297.2023.2190148 · Cited by 3,087
On the Opportunities and Risks of Foundation Models
Bommasani, R., Hudson, D. A., Adeli, E., Altman, R., Arora, S., von Arx, S., Bernstein, M. S., Bohg, J., Bosselut, A., Brunskill, E., et al. (2021) · arXiv · doi:10.48550/arxiv.2108.07258 · Cited by 2,676
Hutcheon, L. (2012) · doi:10.4324/9780203095010 · Cited by 2,292
ChatGPT: Bullshit spewer or the end of traditional assessments in higher education?
(2023) · Journal of Applied Learning & Teaching · doi:10.37074/jalt.2023.6.1.9 · Cited by 1,806
Simple and Controllable Music Generation
Copet, J., Kreuk, F., Gat, I., Remez, T., Kant, D., Synnaeve, G., Adi, Y., & Défossez, A. (2023) · neural information processing systems · doi:10.52202/075280-2066 · Cited by 1,178
Shneiderman, B. (2022) · doi:10.1093/oso/9780192845290.001.0001 · Cited by 955
Kress, G. (2009) · Routledge eBooks · doi:10.4324/9780203970034 · Cited by 763
Research works printed but not linked: 9
- Arora A. (2024). SongGen: Framework for controllable AI song generation through interactive songwriting and artist emulation [Master’s thesis]. Massachusetts Institute of Technology. https://hdl.handle.net/1721.1/157249
- Coleridge S. T. (2004). Christabel (F. H. Sykes Ed.). Project Gutenberg. https://www.gutenberg.org/cache/epub/11101/pg11101-images.html
- Ellis, P. A. (2007). Unity in diversity: Fungi from Yuggoth as a unified setting. Lovecraft Annual, 1, 84–90.
- Glass G. (2024). Fidelity remix and the adaptive potential: Leveraging AI and ML techniques in literary adaptation theory [PhD dissertation]. University of North Carolina at Chapel Hill. Carolina Digital Repository. https://doi.org/10.17615/pkpv-p313 · DOI
- Jewitt, C. (2014). The Routledge handbook of multimodal analysis (2nd ed.). Routledge. https://www.routledge.com/The-Routledge-Handbook-of-Multimodal-Analysis/Jewitt/p/book/9781138245198
- Joshi, S. T., & Schultz, D. E. (2001). An H.P. Lovecraft encyclopedia. Greenwood Press.
- Newman, M., Morris, L., & Lee, J. H. (2023, November 5–9). Human-AI music creation: Understanding the perceptions and experiences of music creators for ethical and productive collaboration [Paper presentation]. Proceedings of the 24th International Society for Music Information Retrieval Conference, Milan, Italy.
- Reynolds, S. (2005). Rip it up and start again: Postpunk 1978–1984. Faber & Faber.
- Yin, R. K. (2018). Case study research and applications: Design and methods (6th ed.). SAGE. https://uk.sagepub.com/en-gb/eur/case-study-research-and-applications/book250150
Metadata
- Authors
- Fotaris, P.
- Affiliations
- University of Brighton (GB)
- Corresponding author
- Fotaris, Panagiotis
- ORCID
- Fotaris, Panagiotis
- Venue
- Cogent Arts & Humanities
- Publisher
- Informa UK Limited
- Published
- 2026-08-27
- Volume
- 13
- Issue
- 1
- ISSN
- 2331-1983 (ISSN-L)
- OpenAlex
- W7204496521
- Cited by
- 0
- References
- 32
- FWCI
- 0.00
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0/ (vor)
- doi.org/10.1080/policies-text-data-mining (tdm)
- Locations
- Cogent Arts and Humanities (publishedVersion, cc-by)
- University of Brighton Repository (University of Brighton) (acceptedVersion, cc-by-nc-nd)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
- Topics
- Social Sciences › Arts and Humanities › Music › Music Technology and Sound Studies
- Social Sciences › Arts and Humanities › Literature and Literary Theory › Digital Humanities and Scholarship
- Social Sciences › Arts and Humanities › Visual Arts and Performing Arts › Artistic and Creative Research
- Keywords
- generative AI, text-to-audio generation, H. P. Lovecraft, literary interpretation, post-punk, indie rock, source selection, post-production
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: AI-assisted literary-to-song adaptation: a practice-based study of literary interpretation in a text-to-audio workflow. https://app.cochat.ai/papers/10.1080/23311983.2026.2722084
- The paper
- Fotaris, P. (2026). AI-assisted literary-to-song adaptation: a practice-based study of literary interpretation in a text-to-audio workflow. Cogent Arts & Humanities. https://doi.org/10.1080/23311983.2026.2722084
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.