Pith. sign in

REVIEW 3 major objections 5 minor 34 references

CSL-L2M: Controllable Song-Level Lyric-to-Melody Generation Based on Conditional Transformer with Fine-Grained Lyric and Musical Controls

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A conditional Transformer generates full-song melodies that align each lyric syllable to one or more notes and follow user-specified musical attributes.

desk verdict REMI-Aligned and the new dataset are solid contributions, but the headline comparison is confounded by target-derived controls and the paper needs a major evaluation overhaul before its claims can be trusted. read the letter →

arxiv 2412.09887 v2 pith:X22IXBWJ submitted 2024-12-13 eess.AS cs.AIcs.SDeess.SP

Li Chai , Donglin Wang This is my paper · ORCID
classification eess.AScs.AIcs.SDeess.SP
keywords lyric-to-melodygenerationcontrollablemusicin-attentionconditioningREMI-AlignedTransformerdecodersyllable-notealignmentChinesepopsongsVQ-VAEmusicalfeatures
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper proposes CSL-L2M, a lyric-to-melody system that aims to generate complete songs rather than short fragments. The central claim is that a Transformer decoder conditioned through in-attention on multi-granularity lyric and musical controls can produce full-song melodies in which every syllable is aligned to one or more notes and user-chosen musical attributes are respected. The paper introduces a new token representation, REMI-Aligned, that makes syllable- and sentence-level lyric-melody alignment explicit, and a dataset of over ten thousand annotated Chinese pop songs. If the claim is right, lyric-based songwriting tools could move from stitching sentence-level melodies into coherent whole songs with controllable key, emotion, structure, and rhythm.

What carries the argument

Two components carry the argument. REMI-Aligned extends the REMI music representation by adding ALIGN tokens at syllable and sentence boundaries, converting the strict but weak lyric-melody correspondence into discrete, learnable tokens that support both one-to-one and one-to-many alignments. In-attention conditioning is the mechanism that makes the controls effective over long sequences: each segment-level control vector is projected into the decoder's hidden-state space and added to the input of every self-attention layer, so the model is continually reminded of the current sentence's lyric and musical conditions. The per-syllable control vector concatenates all lyric and musical embeddings (Eq. 5), which is what lets the same architecture handle both tight alignment and user-directed attribute control.

What would settle it

Run the ablation with the sentence-level statistical controls computed from held-out reference songs or from user-specified values instead of from the ground-truth test melody; if pitch and duration similarity and melody distance collapse toward the no-musical-control baseline, the reported advantage depends on target leakage rather than on learning the lyric-melody mapping.

Watch

Extended reading notes

Core claim

The discovery the paper is trying to establish is that the long-standing weaknesses of lyric-to-melody generation - weak controllability, low quality, and poor structure - can be addressed simultaneously at the full-song level by a single conditional autoregressive Transformer. On the lyric side, the model consumes syllable-level tone embeddings, word-level part-of-speech embeddings, and sentence-level semantic embeddings; on the musical side, it consumes human-labeled tags (key, emotion, structure), sentence-level statistical attributes (pitch, duration, note density, alignment ratio), and learned VQ-VAE features. These are concatenated per syllable and injected into the decoder through in-attention, so the conditions are present at every layer and every time step. The paper reports that on its collected dataset of 10,170 Chinese pop songs, CSL-L2M outperforms prior state-of-the-art models on objective similarity and subjective listening scores, and that ablating control groups progressively degrades performance.

Load-bearing premise

The evaluation assumes that conditioning the decoder on sentence-level statistical attributes extracted from the ground-truth test melody is a fair measure of generation quality, rather than a shortcut that lets the model reproduce the target's summary statistics and inflate similarity scores.

Editorial extensions

If this is right

  • Users could specify key, emotion, and song structure at the whole-song level and fine-grained sentence-level statistics such as pitch mean or note density, and receive melodies that track those specifications.
  • Syllables that naturally span multiple notes are no longer forced into a one-to-one mapping, so melismatic vocal writing can be generated directly rather than avoided.
  • Full-song structure, including repetition patterns and verse-chorus contrast, can emerge from a single decoding pass instead of concatenated sentence-level fragments.
  • The progressive ablation results indicate that each control group - duration, pitch, rhythm, alignment, and global tags - contributes independent information to melody quality, so controllability and quality are not in tension.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • One implication the paper leaves implicit is that a user-facing deployment must source the sentence-level statistical controls without access to the target melody; if users cannot supply them, the controls would need to be predicted or estimated.
  • Another open question is whether the alignment-token representation transfers to other tonal languages, where tone embeddings might similarly guide pitch, or to non-tonal languages, where the tone channel would be dropped.
  • Because the dataset is limited to Chinese pop songs in 4/4 time, the method's generality to other genres, meters, and languages is untested; adapting REMI-Aligned and the control stack to those settings would be a natural next experiment.
  • Running the same Spearman rank-correlation controllability analysis on prior baselines would show whether the fine-grained control this paper claims is a new capability or simply a more favorable evaluation protocol.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes CSL-L2M, a controllable full-song lyric-to-melody generation system based on a Transformer decoder with in-attention conditioning. The method introduces a REMI-Aligned music representation that encodes syllable- and sentence-level lyric-melody alignments, and it feeds a multi-granularity set of lyric controls (tone, part-of-speech, sentence-level semantic embeddings) and musical controls (human-labeled tags, 12 sentence-level statistical musical attributes, and learned VQ-VAE features) into the decoder. The authors present a new dataset of 10,170 Chinese pop songs with alignment annotations and report objective and subjective evaluations comparing CSL-L2M against TeleMelody and SongComposer, along with ablations and a controllability study.

Significance. If the reported results are credible, the paper makes several worthwhile contributions: a full-song lyric-to-melody formulation rather than sentence-level concatenation, explicit support for one-to-many syllable-note alignment through a new representation, a broad set of fine-grained control mechanisms, and a newly collected aligned dataset. The in-attention conditioning architecture and the use of a pre-trained VQ-VAE for high-fidelity musical features are interesting ideas, and the public availability of code and demos is a strength. However, the central claim of outperforming state-of-the-art systems is not supported by the reported experiments because the evaluation protocol appears to condition the model on statistical summaries of the ground-truth test melody, which can trivially inflate the objective similarity metrics. The controllability study using Spearman correlations is a good idea and partially demonstrates attribute control, but it does not resolve the circularity in the main comparison.

major comments (3)
  1. [§Experimental Settings, Table 1] The authors should clarify whether the baseline models were trained or fine-tuned on the same 10,170-song dataset and the same train/validation/test split. The paper states only that 'we first compare our CSL-L2M with two state-of-the-art models, i.e. TeleMelody and SongComposer' without describing whether these models were retrained on the new dataset, whether they use the same full-song generation setting, or whether the same test songs are used. If the baselines were used off-the-shelf without adapting to the new data, differences in training data domain and scale could explain much of the gap in Table 1. Please document the baseline setup in detail or retrain the baselines on the same data and report the results.
  2. [§Subjective Metrics, Table 1] The subjective evaluation is not sufficiently controlled or reported. The paper mentions 10 test songs and 70 participants but provides no inter-annotator agreement, no statistical significance tests, and no description of whether the raters were blinded to the system or to the conditioning information. Since the CSL-L2M outputs evaluated in Table 1 may have been conditioned on ground-truth musical attributes, the subjective ratings could reflect similarity to the original song rather than independent musical quality. A controlled listening test with matched generative conditions and appropriate statistical analysis is needed to support the subjective quality claim.
  3. [§Controllability Study, Figure 5] The controllability study is informative but incomplete. The paper reports Spearman correlations between user-specified attribute classes and attribute values computed from generated melodies, which is a reasonable approach. However, the analysis is only performed for CSL-L2M and does not compare against the controllability of the baseline methods. To support the claim of 'better controllability' over state-of-the-art models, the same Spearman-based controllability evaluation should be applied to the baselines. Additionally, the paper does not specify how many generated samples are used for the correlations or whether the results are averaged over repeated generations.
minor comments (5)
  1. [Throughout] There are several typographical errors, including 'V AE' instead of 'VAE' and 'vise versa' instead of 'vice versa'.
  2. [§Statistical Musical Attributes] The definitions of several statistical attributes (DMM, AA, CM) are not given in the paper; the reader is referred to a website. For an archival publication, these definitions should be stated in the text or in an appendix.
  3. [§Main Results, Table 2] Figure 4 shows a 'w/o all Ctls' setting, but Table 2 does not include that row. Please add the corresponding results to Table 2 or explain the omission.
  4. [§Controllability Study, Figure 5] The description of the correlation matrix in Figure 5 would benefit from a precise explanation of how 'unrelated attribute' correlations are computed and whether they control for the other specified attributes.
  5. [§Dataset] The new dataset is a valuable contribution, but it is not released along with the code. The paper should state whether the dataset will be made publicly available for reproducibility.

Circularity Check

1 steps flagged · score 6.0 of 10

Objective comparison is partly forced by target-derived statistical controls; the reported PD/DD/MD superiority over baselines is confounded by conditioning on ground-truth melody summaries.

  1. fitted input called prediction [Methodology, Eq. (5); Experiments, Objective Metrics; Table 2]
    "cs = concat([clyric_s ; ckey_s ; cemot_s ; cstruc_s ; cPM_s ; cPV_s ; cPR_s ; cDMM_s ; cAA_s ; cCM_s ; cDM_s ; cDV_s ; cDR_s ; cMCD_s ; cND_s ; cAlign_s ; zlearned_s]), yt = Dec(x<t; cs). ... We focus on assessing the similarity between the generated and the ground-truth melodies."

    The decoder is conditioned, at every syllable, on sentence-level statistical attributes (PM, PV, PR, DMM, AA, CM, DM, DV, DR, MCD, ND, Align) that the paper defines as calculated from the melody sequence itself. The objective evaluation measures PD/DD/MD similarity between the generated and ground-truth test melodies. If the statistical controls are set to the ground-truth sentence-level statistics for each test song, as the natural protocol for a controllable-generation comparison against ground truth implies, then the model is asked to emit a melody whose pitch/duration distributions, note density, and alignment ratio match the very summaries it was given.

full rationale

The main load-bearing comparison in Table 1 is confounded: CSL-L2M is evaluated with its sentence-level statistical musical controls, and those controls are defined as statistics of the melody being generated. Since PD/DD/MD score similarity to the ground-truth test melody, using the ground-truth statistics as the user-specified controls makes the objective metrics measure how faithfully the decoder echoes its own input summaries. That is a partial reduction of the claimed prediction to the conditioning input, and it explains the large gap over baselines and the monotone ablation in Table 2. This is not a fully circular derivation: the lyric controls (tone, POS, semantics), REMI-Aligned representation, and the in-attention mechanism are independently motivated, and the subjective evaluation and Spearman controllability analysis are not forced by target leakage. There are no load-bearing self-citations (the authors cite Wu and Yang 2023, von Rütte et al. 2023, etc., but not their own prior work), so no self-citation pattern applies. The score is 6 rather than 8 because the circularity is confined to the objective evaluation protocol; the method itself still has independent components and the controllability claims are meaningful.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the accuracy of the alignment annotations, the sufficiency of the chosen statistical attributes as controllable descriptors, the validity of borrowing in-attention and VQ-VAE features, and the correctness of the Chinese linguistic preprocessing. No new physical or conceptual entities are introduced; REMI-Aligned is a data representation rather than an invented entity with independent evidence.

free parameters (4)
  • Statistical attribute quantization bin count K = 64
    Each of the 12 sentence-level statistical attributes is quantized into K=64 bins with approximately equal sample sizes; the bin boundaries are fit to the training data and control the granularity of all statistical musical controls.
  • VQ-VAE latent group and codebook sizes = 64 groups, 2048 codes
    Hyperparameters of the vector quantizer used to extract learned musical features; chosen by hand and affect the fidelity of the style-transfer control signal.
  • Nucleus sampling temperature and top-p = tau=1.2, p=0.9
    Inference-time sampling hyperparameters selected by the authors; they trade off diversity and quality and influence all reported generation results.
  • Model capacity settings = 12 layers, 8 heads, hidden 512, FFN 2048
    Hand-selected architecture dimensions for the encoder, decoder, and VQ-VAE; these are standard choices but still free parameters that affect performance.
assumptions (5)
  • domain assumption The lyric-melody alignment annotations are accurate, and the REMI-Aligned token sequence faithfully represents the true alignment of each song.
    The whole method relies on strict syllable- and sentence-level alignment being known and correctly encoded in the token stream. This is introduced in the REMI-Aligned Representation section.
  • domain assumption The 12 selected statistical musical attributes capture the salient controllable dimensions of melody.
    The paper assumes that pitch mean/variance/range, duration statistics, note density, and alignment ratio are sufficient to steer generated melodies toward user intent. This is introduced in the Statistical Musical Attributes section.
  • domain assumption In-attention conditioning from MuseMorphose transfers successfully from piano style transfer to lyric-to-melody generation.
    The paper adopts Eqs. (2) and (3) from Wu and Yang 2023 without re-deriving them, assuming the same tight-control mechanism works for full-song melody generation conditioned on lyrics and musical attributes.
  • ad hoc to paper The pre-trained VQ-VAE latent features provide high-fidelity melody information without leaking the exact token sequence.
    The VQ-VAE is trained on the same corpus, and Table 2 shows the with-learned-controls version reaches MD 0.25, indicating near-reconstruction. The paper uses these features as controls while acknowledging they carry target-melody information.
  • domain assumption Jieba POS tagging and Mandarin tone labels are correct for the Chinese lyrics.
    The tone and POS embeddings are treated as reliable lyrical features; errors in these external tools would propagate into the conditioning signals.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CSL-L2M: Controllable Song-Level Lyric-to-Melody Generation Based on Conditional Transformer with Fine-Grained Lyric and Musical Controls." pith.science (2026). https://pith.science/paper/X22IXBWJ

@misc{pith2026241209887,
  author       = {Pith},
  title        = {Pith review of: CSL-L2M: Controllable Song-Level Lyric-to-Melody Generation Based on Conditional Transformer with Fine-Grained Lyric and Musical Controls},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/X22IXBWJ}},
  note         = {Machine review of arXiv:2412.09887}
}
read the original abstract

Lyric-to-melody generation is a highly challenging task in the field of AI music generation. Due to the difficulty of learning strict yet weak correlations between lyrics and melodies, previous methods have suffered from weak controllability, low-quality and poorly structured generation. To address these challenges, we propose CSL-L2M, a controllable song-level lyric-to-melody generation method based on an in-attention Transformer decoder with fine-grained lyric and musical controls, which is able to generate full-song melodies matched with the given lyrics and user-specified musical attributes. Specifically, we first introduce REMI-Aligned, a novel music representation that incorporates strict syllable- and sentence-level alignments between lyrics and melodies, facilitating precise alignment modeling. Subsequently, sentence-level semantic lyric embeddings independently extracted from a sentence-wise Transformer encoder are combined with word-level part-of-speech embeddings and syllable-level tone embeddings as fine-grained controls to enhance the controllability of lyrics over melody generation. Then we introduce human-labeled musical tags, sentence-level statistical musical attributes, and learned musical features extracted from a pre-trained VQ-VAE as coarse-grained, fine-grained and high-fidelity controls, respectively, to the generation process, thereby enabling user control over melody generation. Finally, an in-attention Transformer decoder technique is leveraged to exert fine-grained control over the full-song melody generation with the aforementioned lyric and musical conditions. Experimental results demonstrate that our proposed CSL-L2M outperforms the state-of-the-art models, generating melodies with higher quality, better controllability and enhanced structure. Demos and source code are available at https://lichaiustc.github.io/CSL-L2M/.

Figures

Figures reproduced from arXiv: 2412.09887 by the authors.

Figure 1
Figure 1. Two representations of the same music piece. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Architecture of CSL-L2M. In contrast to English, Chinese is a tonal language contain￾ing four main tones and one light tone in its characters. The pitch flow of a melody in Chinese songs is usually closely re￾lated to the tones of the corresponding lyrics. Accordingly, we incorporate the tone information into our model train￾ing to help learn pitch flow of the generated melodies to match with the given lyrics. The s… view at source ↗
Figure 3
Figure 3. Distributions of music attributes in our paired [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Training dynamics of our CSL-L2M under differ [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Spearman’s rank correlation coefficients between [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Case study on generated melodies from CSL-L2M. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 25 canonical work pages

  1. [1]

    Bao, H.; Huang, S.; Wei, F.; Cui, L.; Wu, Y.; Tan, C.; Piao, S.; and Zhou, M. 2019. Neural melody composition from lyrics. In Natural Language Processing and Chinese Computing: 8th CCF International Conference, NLPCC 2019, Dunhuang, China, October 9--14, 2019, Proceedings, Part I 8, 499--511. Springer

    work page 2019

  2. [2]

    Briot, J.-P.; and Pachet, F. 2020. Deep learning for music generation: challenges and directions. Neural Computing and Applications, 32(4): 981--993

    work page 2020

  3. [3]

    Brunner, G.; Konrad, A.; Wang, Y.; and Wattenhofer, R. 2018. MIDI-VAE: Modeling dynamics and instrumentation of music with applications to style transfer. In ISMIR, 747--754

    work page 2018

  4. [4]

    Choi, K.; Hawthorne, C.; Simon, I.; Dinculescu, M.; and Engel, J. 2020. Encoding musical style with transformer autoencoders. In International conference on machine learning, 1899--1908. PMLR

    work page 2020

  5. [5]

    Ding, S.; Liu, Z.; Dong, X.; Zhang, P.; Qian, R.; He, C.; Lin, D.; and Wang, J. 2024. Songcomposer: A large language model for lyric and melody composition in song generation. arXiv preprint arXiv:2402.17645

    arXiv 2024

  6. [6]

    Dong, H.-W.; Hsiao, W.-Y.; Yang, L.-C.; and Yang, Y.-H. 2018. Musegan: Multi-track sequential generative adversarial networks for symbolic music generation and accompaniment. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 32

    work page 2018

  7. [7]

    Duan, W.; Yu, Y.; and Oyama, K. 2024. Semantic dependency network for lyrics generation from melody. Neural Computing and Applications, 36(8): 4059--4069

    work page 2024

  8. [8]

    Duan, W.; Zhang, Z.; Yu, Y.; and Oyama, K. 2022. Interpretable melody generation from lyrics with discrete-valued adversarial training. In Proceedings of the 30th ACM international conference on multimedia, 6973--6975

    work page 2022

Show all 34 references
  1. [9]

    Hahn, S.; Zhu, R.; Mak, S.; Rudin, C.; and Jiang, Y. 2023. An Interpretable, Flexible, and Interactive Probabilistic Framework for Melody Generation. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 4089--4099

    2023

  2. [10]

    Holtzman, A.; Buys, J.; Du, L.; Forbes, M.; and Choi, Y. 2020. The curious case of neural text degeneration. The Eighth International Conference on Learning Representations

    2020

  3. [11]

    Huang, Y.-S.; and Yang, Y.-H. 2020. Pop music transformer: Beat-based modeling and generation of expressive pop piano compositions. In Proceedings of the 28th ACM international conference on multimedia, 1180--1188

    2020

  4. [12]

    Ju, Z.; Lu, P.; Tan, X.; Wang, R.; Zhang, C.; Wu, S.; Zhang, K.; Li, X.; Qin, T.; and Liu, T.-Y. 2021. Telemelody: Lyric-to-melody generation with a template-based two-stage method. arXiv preprint arXiv:2109.09617

    2021 arXiv

  5. [13]

    Kawai, L.; Esling, P.; and Harada, T. 2020. Attributes-Aware Deep Music Transformation. In ISMIR, 670--677

    2020

  6. [14]

    C.-W.; and Sze, R

    Long, C.; Wong, R. C.-W.; and Sze, R. K. W. 2013. T-music: A melody composer based on frequent pattern mining. In 2013 IEEE 29th International Conference on Data Engineering (ICDE), 1332--1335. IEEE

    2013

  7. [15]

    R.; and Ventura, D

    Monteith, K.; Martinez, T. R.; and Ventura, D. 2012. Automatic Generation of Melodic Accompaniments for Lyrics. In ICCC, 87--94

    2012

  8. [16]

    Neves, P.; Fornari, J.; and Florindo, J. 2022. Generating music with sentiment using Transformer-GANs. In ISMIR, 717--725

    2022

  9. [17]

    Nichols, E. 2009. Lyric-based rhythm suggestion. In ICMC

    2009

  10. [18]

    Payne, C. 2019. MuseNet. https://openai.com/blog/musenet

    2019

  11. [19]

    Roberts, A.; Engel, J.; Raffel, C.; Hawthorne, C.; and Eck, D. 2018. A hierarchical latent vector model for learning long-term structure in music. In International conference on machine learning, 4364--4373. PMLR

    2018

  12. [20]

    Sarmento, P.; Kumar, A.; Chen, Y.-H.; Carr, C.; Zukowski, Z.; and Barthet, M. 2023. GTR-CTRL: instrument and genre conditioning for guitar-focused music generation with transformers. In International Conference on Computational Intelligence in Music, Sound, Art and Design (Par...

    2023

  13. [21]

    Sheng, Z.; Song, K.; Tan, X.; Ren, Y.; Ye, W.; Zhang, S.; and Qin, T. 2021. Songmass: Automatic song writing with pre-training and alignment constraint. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, 13798--13805

    2021

  14. [22]

    R.; Wu, J.; Tang, S.; Li, W.; and Yu, Y

    Srivastava, A.; Duan, W.; Shah, R. R.; Wu, J.; Tang, S.; Li, W.; and Yu, Y. 2022. Melody generation from lyrics using three branch conditional LSTM-GAN. In International Conference on Multimedia Modeling, 569--581. Springer

    2022

  15. [23]

    H.; and Herremans, D

    Tan, H. H.; and Herremans, D. 2020. Music FaderNets: Controllable Music Generation Based On High-Level Features via Low-Level Feature Modelling. In ISMIR, 109--116

    2020

  16. [24]

    Tian, Y.; Narayan-Chen, A.; Oraby, S.; Cervone, A.; Sigurdsson, G.; Tao, C.; Zhao, W.; Chen, Y.; Chung, T.; Huang, J.; et al. 2023. Unsupervised melody-to-lyric generation. arXiv preprint arXiv:2305.19228

    2023 arXiv

  17. [25]

    Touvron, H.; Martin, L.; Stone, K.; Albert, P.; Almahairi, A.; Babaei, Y.; Bashlykov, N.; Batra, S.; Bhargava, P.; Bhosale, S.; et al. 2023. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288

    2023 arXiv

  18. [26]

    N.; Kaiser, .; and Polosukhin, I

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, .; and Polosukhin, I. 2017. Attention is all you need. Advances in neural information processing systems, 30

    2017

  19. [27]

    von R \"u tte, D.; Biggio, L.; Kilcher, Y.; and Hofmann, T. 2023. FIGARO: Controllable music generation using learned and expert features. In The Eleventh International Conference on Learning Representations

    2023

  20. [28]

    Wu, S.-L.; and Yang, Y.-H. 2023. MuseMorphose: Full-song and fine-grained piano music style transfer with one transformer VAE. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 31: 1953--1967

    2023

  21. [29]

    Yang, L.-C.; Chou, S.-Y.; and Yang, Y.-H. 2017. MidiNet: A convolutional generative adversarial network for symbolic-domain music generation. In ISMIR, 324--331

    2017

  22. [30]

    Yu, Y.; Srivastava, A.; and Canales, S. 2021. Conditional LSTM-GAN for melody generation from lyrics. ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM), 17(1): 1--20

    2021

  23. [31]

    Zhang, C.; Chang, L.; Wu, S.; Tan, X.; Qin, T.; Liu, T.-Y.; and Zhang, K. 2022. Relyme: improving lyric-to-melody generation by incorporating lyric-melody relationships. In Proceedings of the 30th ACM International Conference on Multimedia, 1047--1056

    2022

  24. [32]

    Zhang, Z.; Yu, Y.; and Takasu, A. 2023. Controllable lyrics-to-melody generation. Neural Computing and Applications, 35(27): 19805--19819

    2023

  25. [33]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  26. [34]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.