Everyone-Can-Sing: Zero-Shot Singing Voice Synthesis and Conversion with Speech Reference

Roger B. Dannenberg; Shuqi Dai; Yunyun Wang; Zeyu Jin

REVIEW 3 major objections 5 minor 25 references

Everyone-Can-Sing: Zero-Shot Singing Voice Synthesis and Conversion with Speech Reference

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A five-second speech sample is enough to synthesize new singing or convert existing singing in an unseen voice.

desk verdict A solid, honest zero-shot SVS/SVC system paper whose main weaknesses are weak baselines and an untested-but-disclosed timbre premise, not the architecture. read the letter →

arxiv 2501.13870 v1 pith:UXKKMTXM submitted 2025-01-23 cs.SD eess.AS

Shuqi Dai , Yunyun Wang , Roger B. Dannenberg , Zeyu Jin This is my paper · ORCID

classification cs.SDeess.AS

keywords singingvoicesynthesisconversionzero-shotlearningtimbretransferspeechreferencediffusionmodelmixedspeech-singingtrainingexpressiveperformancecontrol

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

The paper proposes a unified framework that turns a short speech recording into a voice identity for singing. It claims that one zero-shot singing-voice synthesizer and two zero-shot singing-voice converters can all be driven by a five-second speech reference, without any singing recording from that person. The framework separates singing into performance attributes, pronunciation content, and vocal timbre, then recombines them with a diffusion-based acoustic model trained on both singing and speech data. The authors' experiments report that their zero-shot synthesizer matches an unseen speech reference's timbre better than existing zero-shot systems, while keeping singing quality comparable to a non-zero-shot baseline. If true, this would remove a major data bottleneck in singing-voice cloning: a brief spoken sample would suffice as the voice condition.

What carries the argument

The load-bearing object is a three-way decomposition of a singing signal: expressive performance attributes (F0 contour, amplitude envelope, timing), a pronunciation/content embedding derived from lyrics or from a self-supervised speech content encoder, and a time-independent timbre embedding from a pre-trained speaker-embedding model. These conditions are fed to a diffusion-based acoustic model that generates a mel-spectrogram, then a neural vocoder turns it into a waveform. The mechanism enabling zero-shot transfer is the swappability of the timbre embedding: at inference, the training speaker identity is replaced by the embedding of the unseen five-second speech reference, while F0 and amplitude are either generated from the score or extracted from the input singing. Mixed training on singing plus a large multi-speaker speech corpus exposes the model to far more phonetic and prosodic variety than singing data alone, which the paper argues mitigates singing-data scarcity.

What would settle it

Take a set of speakers whose spoken and sung voices are very different (for example, operatic singers with strong vibrato and head voice), compute the embedding distance between each speaker's short speech sample and their own singing; if those same-speaker speech-to-singing distances are no smaller than distances between different speakers, then the speech embedding does not contain the timbre dimensions needed for singing, and the reported cross-domain similarity should fail for such voices.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that cross-domain zero-shot singing synthesis and conversion can be achieved with a five-second speech reference by explicitly disentangling voice, content, and expressive performance, and by training the synthesis backbone on a mixture of singing and speech data. The SVS model takes a musical score, lyrics, a style token, and a speech reference; the two SVC models take an existing singing sample plus either aligned lyrics or a pre-trained content embedding, with the same kind of speech reference. Subjective listening tests show all three models producing singing that is rated close in quality to the non-zero-shot score-based baseline, while the zero-shot SVS model receives the highest timbre-similarity rating among the zero-shot systems tested. The authors interpret this as evidence that a speech-trained speaker embedding can carry enough timbre information to condition singing, and that expanding singing training with speech data improves both quality and similarity.

Load-bearing premise

The claim rests on the premise that a speech-derived speaker embedding contains enough singing-relevant timbre information to transfer a person's voice into a song.

Editorial extensions

If this is right

A five-second speech recording can serve as the voice condition for generating a new song in that person's singing voice, without any sung samples from them.
Existing singing can be converted to an unseen speaker's timbre using either aligned lyrics or only the audio itself, since a pre-trained content embedding can stand in for lyrics.
Speech data can be added to singing-model training at a 1:1 ratio without hurting quality, which lowers the data requirement for expressive singing synthesis.
Shifting the target score into the speech reference's pitch range improves timbre similarity, though it can lower perceived musicality when the shift is large.
The same separation of content, performance, and timbre is proposed as a template for other low-data music tasks such as instrumental style transfer.

Reading between the lines

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

An implication the paper leaves implicit: the same speech-reference conditioning should work with references longer than five seconds, and similarity ratings would likely rise until the added speech introduces pronunciation habits that conflict with the singing content.
A testable extension would be to replace the speech-trained speaker embedding with one trained on both speech and singing (or on singing alone) and compare similarity on singers whose spoken and sung timbres differ substantially, such as operatic voices.
The framework's separation of performance attributes from timbre suggests that instrumental timbre transfer could be attempted by conditioning the acoustic model on instrument embeddings instead of voice embeddings, using the same diffusion backbone.
Because the two SVC models differ mainly in whether pronunciation comes from aligned lyrics or a self-supervised content encoder, the gap between them hints that content disentanglement quality, not score information, is the next bottleneck for conversion.

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.

Desk Editor's Note

read the letter

You should know this paper is a competent engineering contribution, not a paradigm shift. The genuinely new parts are the unified framework that does zero-shot SVS and two flavors of SVC from a 5-second speech reference, and the mixed speech-singing training recipe that helps with data scarcity. The authors reuse ExpressiveSinger, GR0, Resemblyzer, and BigVGAN, and they say so plainly. That is fine; the contribution is in the integration and the evaluation, not in new building blocks.

The evaluation is the strongest part. They ran a large subjective test with 487 valid surveys, covering quality and similarity, with multiple ablations. They also report honestly where things fail: singing references beat speech references on similarity, opera is harder than pop, and pitch adjustment can hurt quality. That transparency earns credibility.

The soft spots are real but addressable. The zero-shot SVS baseline is ExpressiveSinger, which is not zero-shot, so the headline "best among zero-shot approaches" is supported only by comparison against GR0 and the four-demo Unified baseline. That is thin. There is no code or data release, which limits independent verification. The deeper concern raised about Resemblyzer is legitimate but not fatal: the paper never directly shows the speech embedding captures vibrato, head voice, or high-range projection, and its own ablation shows a consistent similarity gap between singing and speech references. But the paper does not overclaim—it frames the speech-reference similarity as an improvement over baselines, and the SIM-MOS scores around 3.5–3.8 suggest the embedding transfers at least coarse identity. The stress-test note is correct that this is an untested premise; I would call it a limitation, not a load-bearing flaw, because the model's outputs are judged against the actual speech reference and they do score above mismatched pairs.

Who gets value from this: anyone building SVS/SVC systems, especially in low-resource or cross-domain settings, and anyone designing subjective evaluations for generative audio. It deserves a serious referee. I would send it out, ask the authors to strengthen the zero-shot baseline comparison and to release demo code or at least the trained models, and then accept.

Recommendation: send to peer review with requests for stronger baselines and reproducible artifacts.

Referee Report

3 major / 5 minor

Summary. The manuscript proposes a unified zero-shot framework for singing voice synthesis (SVS) and two variants of singing voice conversion (SVC), all conditioned on a five-second speech reference for voice identity. The system extends ExpressiveSinger with a Resemblyzer speaker embedding, uses a diffusion-based acoustic model, and integrates GR0's content embedding for one SVC variant. Training incorporates mixed singing and speech data. The paper reports a large subjective evaluation with 487 valid surveys, including a comparison study and an ablation study, and claims substantial improvements in timbre similarity and musicality over existing baselines.

Significance. If the central claim holds, the framework demonstrates that a short speech sample can serve as a sufficient voice-identity condition for high-quality singing synthesis and conversion, which is practically valuable given the scarcity of singing data. The paper's strengths include a large and carefully filtered subjective evaluation, a broad ablation study covering reference type, pitch adjustment, cross-language and cross-gender conditions, and a design that reuses pretrained components to reduce training burden. The main limitations are the weak baseline comparison set and an untested assumption about the completeness of the speech-derived timbre embedding for singing.

major comments (3)

[III-B, Table I] The claim that the proposed zero-shot SVS 'delivers the highest similarity among all zero-shot approaches' is not well supported by the baseline set. ExpressiveSinger is not a zero-shot model and uses a same-singer singing reference from the training data, so its SIM-MOS of 4.24 is an upper bound rather than a comparable score. The Unified baseline is represented by only four demo pairs, as the authors acknowledge, which is insufficient for a reliable MOS comparison. GR0 is a speech conversion model rather than a singing synthesis or conversion model. The table therefore demonstrates superiority only over weak or non-comparable baselines; a genuine zero-shot SVS baseline with a comparable number of demos should be included, or the claims should be qualified accordingly.
[II-B and Table II] The central premise that a speech-trained Resemblyzer embedding fully specifies singing timbre is not directly tested. The ablation in Table II shows a consistent similarity gap between singing and speech references across all three models (e.g., zero-shot SVS SIM-MOS 3.84 versus 3.50, SVC(b) 3.66 versus 3.47, and SVC(c) 3.76 versus 3.50). This gap is consistent with the speech embedding lacking singing-specific dimensions such as vibrato, head voice, and high-range projection. To support the zero-shot speech-reference claim, the paper should add a speaker-attribution test, a same-speaker speech-to-singing comparison, or an analysis of the embedding's coverage of singing-relevant timbre dimensions.
[II-D] The paper states that GR0's content encoder 'does not include voice timbre information' and uses this as a basis for the SVC(c) model, but this property is inherited from GR0's training on speech and is not verified for singing inputs. Since the acoustic model is trained on singing data with this encoder, possible timbre leakage from the content embedding could affect conversion quality or similarity. Adding an objective content-preservation metric, such as phone accuracy or F0 correlation between the input and converted singing, would strengthen the SVC(c) results.

minor comments (5)

[II-D] The sentence 'keeping all other training settings except the training' appears truncated; please clarify what is held constant versus what changes.
[Table I] The notation 'significance levels < 0.001' is not accompanied by a description of the statistical test used; please specify the test and how the confidence intervals were computed.
[Table II] The header '”Original Pitch”' contains mismatched quotation marks; please correct the typography.
[II-B] The addition of a Leaky ReLU layer to the pronunciation content encoder is mentioned but not evaluated in the ablations; a brief explanation of its effect would help the reader understand its role.
[I, contribution (5)] The paper claims its experiments 'also offer insights for speech,' but no speech-domain experiments are reported; either add such evidence or remove this claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the system combines pre-existing components and its headline results are independent human-MOS evaluations, not consequences of the reused self-cited modules.

full rationale

The paper is empirical system-building rather than a derivation. The zero-shot SVS and SVC models condition a diffusion acoustic model on a pre-trained speech embedding (Resemblyzer) and pre-trained content/performance modules (ExpressiveSinger, GR0), then evaluate on held-out speech and singing references with subjective MOS. No equation in the paper defines the predicted similarity as a function of the conditioning embedding, and the evaluation asks human raters to judge timbre similarity, so the reported improvements are not forced by construction. The reuse of ExpressiveSinger and GR0 is self-citation, but these are externally evaluated building blocks with stated assumptions; the paper does not invoke a uniqueness theorem or forbid alternatives, and its own SVC(c) outperforms the GR0 baseline even though it shares GR0's content encoder, showing the comparison is not a tautology. The ablation honestly reports that singing references yield higher similarity than speech references, which is a limitation of the speech-embedding premise rather than a circular step. Therefore the central claim has independent empirical content and no circularity is identified.

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

The framework's central claim rests on several transfer assumptions: disentanglement of voice, content, and style, speech-to-singing timbre transfer via Resemblyzer, timbre-free content from GR0, and the validity of subjective MOS. These are domain assumptions, not standard math axioms. Free parameters are training and inference hyperparameters chosen by hand. No new physical or conceptual entities are introduced.

free parameters (4)

Mixed training ratio = 1:1
The ratio of singing to speech data during mixed training is set by hand; the ablation shows mixed training improves scores, but no sweep over ratios is reported.
Pitch adjustment octave range = within one octave of the speech reference
An inference-time heuristic to handle pitch mismatches between speech and singing; no systematic optimization or sensitivity analysis is provided.
Diffusion hyperparameters = T=1000, beta from 0.0001 to 0.02, learning rate 2e-4, 900k iterations
Training settings chosen without a reported search; the central claim does not depend on exact values, but they are free choices that could affect results.
Style token set = pop/opera, normal/vibrato
The style and technique labels are categorical inputs chosen from the dataset, which limits generalization to other styles and techniques.

assumptions (5)

domain assumption Voice, content, and singing style are naturally disentangled in the signal and can be modeled by separate pre-trained embeddings.
Stated in Section II-A: 'Our model is built on the assumption that voice, content, and singing styles are naturally disentangled.' This is the central modeling premise.
domain assumption Resemblyzer's speaker embedding captures timbre identity transferable from speech to singing.
Section II-B replaces the symbolic Singer ID with Resemblyzer; the paper does not verify the embedding's coverage of singing-relevant timbre dimensions.
domain assumption GR0's content encoder is free of voice timbre information.
Section II-D relies on the GR0 content embedding for pronunciation without timbre; this property is cited from prior work and not re-verified in the singing domain.
domain assumption Subjective MOS ratings are a valid and sufficient measure of singing quality and timbre similarity.
Section III-B uses only subjective evaluation; no objective metrics are reported to corroborate the listening scores.
domain assumption ExpressiveSinger's pre-trained performance modules transfer unchanged to the new zero-shot setting.
Section II-B reuses ExpressiveSinger's timing, F0, and amplitude modules without retraining; any domain shift from the new voice and content conditions is not measured.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Everyone-Can-Sing: Zero-Shot Singing Voice Synthesis and Conversion with Speech Reference." pith.science (2026). https://pith.science/paper/UXKKMTXM

@misc{pith2026250113870,
  author       = {Pith},
  title        = {Pith review of: Everyone-Can-Sing: Zero-Shot Singing Voice Synthesis and Conversion with Speech Reference},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UXKKMTXM}},
  note         = {Machine review of arXiv:2501.13870}
}

read the original abstract

We propose a unified framework for Singing Voice Synthesis (SVS) and Conversion (SVC), addressing the limitations of existing approaches in cross-domain SVS/SVC, poor output musicality, and scarcity of singing data. Our framework enables control over multiple aspects, including language content based on lyrics, performance attributes based on a musical score, singing style and vocal techniques based on a selector, and voice identity based on a speech sample. The proposed zero-shot learning paradigm consists of one SVS model and two SVC models, utilizing pre-trained content embeddings and a diffusion-based generator. The proposed framework is also trained on mixed datasets comprising both singing and speech audio, allowing singing voice cloning based on speech reference. Experiments show substantial improvements in timbre similarity and musicality over state-of-the-art baselines, providing insights into other low-data music tasks such as instrumental style transfer. Examples can be found at: everyone-can-sing.github.io.

Figures

Figures reproduced from arXiv: 2501.13870 by the authors.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 19 canonical work pages

[1]

Fastspeech 2: Fast and high-quality end-to-end text to speech

Yi Ren, Chenxu Hu, Xu Tan, Tao Qin, Sheng Zhao, Zhou Zhao, and Tie-Yan Liu. Fastspeech 2: Fast and high-quality end-to-end text to speech. In International Conference on Learning Representations , 2020

work page 2020
[2]

Hifi- gan: Generative adversarial networks for efficient and high fidelity speech synthesis

Jungil Kong, Jaehyeon Kim, and Jaekyoung Bae. Hifi- gan: Generative adversarial networks for efficient and high fidelity speech synthesis. Advances in Neural Information Processing Systems, 33:17022–17033, 2020

work page 2020
[3]

Bigvgan: A universal neural vocoder with large-scale training

Sang-gil Lee, Wei Ping, Boris Ginsburg, Bryan Catan- zaro, and Sungroh Yoon. Bigvgan: A universal neural vocoder with large-scale training. In The Eleventh International Conference on Learning Representations , 2022

work page 2022
[4]

High-fidelity audio compression with improved rvqgan

Rithesh Kumar, Prem Seetharaman, Alejandro Luebs, Ishaan Kumar, and Kundan Kumar. High-fidelity audio compression with improved rvqgan. Advances in Neural Information Processing Systems , 36, 2024

work page 2024
[5]

Soundstream: An end-to-end neural audio codec

Neil Zeghidour, Alejandro Luebs, Ahmed Omran, Jan Skoglund, and Marco Tagliasacchi. Soundstream: An end-to-end neural audio codec. IEEE/ACM Transactions on Audio, Speech, and Language Processing , 30:495– 507, 2021

work page 2021
[6]

RMSSinger: Realistic-music-score based singing voice synthesis

Jinzheng He, Jinglin Liu, Zhenhui Ye, Rongjie Huang, Chenye Cui, Huadai Liu, and Zhou Zhao. RMSSinger: Realistic-music-score based singing voice synthesis. In Findings of the Association for Computational Linguis- tics: ACL 2023 , pages 236–248, Toronto, Canada, 2023. Association for Computational Linguistics

work page 2023
[7]

Xiaoicesing: A high-quality and integrated singing voice synthesis system

Peiling Lu, Jie Wu, Jian Luan, Xu Tan, and Li Zhou. Xiaoicesing: A high-quality and integrated singing voice synthesis system. arXiv preprint arXiv:2006.06261 , 2020

arXiv 2006
[8]

The singing voice con- version challenge 2023

Wen-Chin Huang, Lester Phillip Violeta, Songxiang Liu, Jiatong Shi, and Tomoki Toda. The singing voice con- version challenge 2023. In 2023 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), pages 1–8. IEEE, 2023

work page 2023

Show all 25 references

[9]

Gr0: Self-supervised global representation learning for zero-shot voice conversion

Yunyun Wang, Jiaqi Su, Adam Finkelstein, and Zeyu Jin. Gr0: Self-supervised global representation learning for zero-shot voice conversion. In ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE, 2024

2024
[10]

Neural analysis and syn- thesis: Reconstructing speech from self-supervised rep- resentations

Hyeong-Seok Choi, Juheon Lee, Wansoo Kim, Jie Lee, Hoon Heo, and Kyogu Lee. Neural analysis and syn- thesis: Reconstructing speech from self-supervised rep- resentations. Advances in Neural Information Processing Systems, 34:16251–16265, 2021

2021
[11]

Neural codec language models are zero-shot text to speech synthesizers

Chengyi Wang, Sanyuan Chen, Yu Wu, Ziqiang Zhang, Long Zhou, Shujie Liu, Zhuo Chen, Yanqing Liu, Huam- ing Wang, Jinyu Li, et al. Neural codec language models are zero-shot text to speech synthesizers. arXiv preprint arXiv:2301.02111, 2023

2023 arXiv
[12]

Nansy++: Unified voice synthesis with neural analysis and synthesis

Hyeong-Seok Choi, Jinhyeok Yang, Juheon Lee, and Hyeongju Kim. Nansy++: Unified voice synthesis with neural analysis and synthesis. arXiv preprint arXiv:2211.09407, 2022

2022 arXiv
[13]

https://github.com/svc-develop-team/so-vits-svc

Softvc vits singing voice conversion. https://github.com/svc-develop-team/so-vits-svc
[14]

Midi-voice: Expressive zero-shot singing voice synthesis via midi-driven priors

Dong-Min Byun, Sang-Hoon Lee, Ji-Sang Hwang, and Seong-Whan Lee. Midi-voice: Expressive zero-shot singing voice synthesis via midi-driven priors. In ICASSP 2024 - 2024 IEEE International Conference on Acous- tics, Speech and Signal Processing (ICASSP) , pages 12622–12626, 2024

2024
[15]

Zero-shot singing voice synthesis from musical score

Jun-You Wang, Hung-Yi Lee, Jyh-Shing Roger Jang, and Li Su. Zero-shot singing voice synthesis from musical score. In 2023 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), pages 1–8, 2023

2023
[16]

Expressivesinger: Multilingual and multi- style score-based singing voice synthesis with expressive performance control

Shuqi Dai, Ming-Yu Liu, Rafael Valle, and Siddharth Gururani. Expressivesinger: Multilingual and multi- style score-based singing voice synthesis with expressive performance control. In ACM Multimedia 2024 , 2024

2024
[17]

Music style transfer: A position paper

Shuqi Dai, Zheng Zhang, and Gus G Xia. Music style transfer: A position paper. In Proceedings of 6th International Workshop on Musical Metacreation , 2018

2018
[18]

A unified model for zero-shot singing voice con- version and synthesis

Jui-Te Wu, Jun-You Wang, Jyh-Shing Roger Jang, and Li Su. A unified model for zero-shot singing voice con- version and synthesis. In Ismir 2022 Hybrid Conference, 2022

2022
[19]

Naturalspeech 2: Latent diffusion models are natural and zero-shot speech and singing synthesizers

Kai Shen, Zeqian Ju, Xu Tan, Yanqing Liu, Yichong Leng, Lei He, Tao Qin, Sheng Zhao, and Jiang Bian. Naturalspeech 2: Latent diffusion models are natural and zero-shot speech and singing synthesizers. arXiv preprint arXiv:2304.09116, 2023

2023 arXiv
[20]

Generalized end-to-end loss for speaker verifi- cation

Li Wan, Quan Wang, Alan Papir, and Ignacio Lopez Moreno. Generalized end-to-end loss for speaker verifi- cation. In 2018 IEEE International Conference on Acous- tics, Speech and Signal Processing (ICASSP) , pages 4879–4883. IEEE, 2018

2018
[21]

Ddsp: Differentiable digital signal processing

Jesse Engel, Chenjie Gu, Adam Roberts, et al. Ddsp: Differentiable digital signal processing. In International Conference on Learning Representations , 2019

2019
[22]

wav2vec 2.0: A framework for self- supervised learning of speech representations

Alexei Baevski, Yuhao Zhou, Abdelrahman Mohamed, and Michael Auli. wav2vec 2.0: A framework for self- supervised learning of speech representations. Advances in neural information processing systems , 33:12449– 12460, 2020

2020
[23]

Dannenberg

Shuqi Dai, Siqi Chen, Yuxuan Wu, Ruxin Diao, Roy Huang, and Roger B. Dannenberg. Singstyle111: A multilingual singing dataset with style transfer. In Proc. of the 24th Int. Society for Music Information Retrieval Conf., 2023

2023
[24]

Libritts-r: A restored multi-speaker text-to-speech corpus

Yuma Koizumi, Heiga Zen, Shigeki Karita, Yifan Ding, Kohei Yatabe, Nobuyuki Morioka, Michiel Bacchiani, Yu Zhang, Wei Han, and Ankur Bapna. Libritts-r: A restored multi-speaker text-to-speech corpus. arXiv preprint arXiv:2305.18802, 2023

2023 arXiv
[25]

Denoising diffusion implicit models

Jiaming Song, Chenlin Meng, and Stefano Ermon. Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502, 2020

2010 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

Fastspeech 2: Fast and high-quality end-to-end text to speech

Yi Ren, Chenxu Hu, Xu Tan, Tao Qin, Sheng Zhao, Zhou Zhao, and Tie-Yan Liu. Fastspeech 2: Fast and high-quality end-to-end text to speech. In International Conference on Learning Representations , 2020

work page 2020

[2] [2]

Hifi- gan: Generative adversarial networks for efficient and high fidelity speech synthesis

Jungil Kong, Jaehyeon Kim, and Jaekyoung Bae. Hifi- gan: Generative adversarial networks for efficient and high fidelity speech synthesis. Advances in Neural Information Processing Systems, 33:17022–17033, 2020

work page 2020

[3] [3]

Bigvgan: A universal neural vocoder with large-scale training

Sang-gil Lee, Wei Ping, Boris Ginsburg, Bryan Catan- zaro, and Sungroh Yoon. Bigvgan: A universal neural vocoder with large-scale training. In The Eleventh International Conference on Learning Representations , 2022

work page 2022

[4] [4]

High-fidelity audio compression with improved rvqgan

Rithesh Kumar, Prem Seetharaman, Alejandro Luebs, Ishaan Kumar, and Kundan Kumar. High-fidelity audio compression with improved rvqgan. Advances in Neural Information Processing Systems , 36, 2024

work page 2024

[5] [5]

Soundstream: An end-to-end neural audio codec

Neil Zeghidour, Alejandro Luebs, Ahmed Omran, Jan Skoglund, and Marco Tagliasacchi. Soundstream: An end-to-end neural audio codec. IEEE/ACM Transactions on Audio, Speech, and Language Processing , 30:495– 507, 2021

work page 2021

[6] [6]

RMSSinger: Realistic-music-score based singing voice synthesis

Jinzheng He, Jinglin Liu, Zhenhui Ye, Rongjie Huang, Chenye Cui, Huadai Liu, and Zhou Zhao. RMSSinger: Realistic-music-score based singing voice synthesis. In Findings of the Association for Computational Linguis- tics: ACL 2023 , pages 236–248, Toronto, Canada, 2023. Association for Computational Linguistics

work page 2023

[7] [7]

Xiaoicesing: A high-quality and integrated singing voice synthesis system

Peiling Lu, Jie Wu, Jian Luan, Xu Tan, and Li Zhou. Xiaoicesing: A high-quality and integrated singing voice synthesis system. arXiv preprint arXiv:2006.06261 , 2020

arXiv 2006

[8] [8]

The singing voice con- version challenge 2023

Wen-Chin Huang, Lester Phillip Violeta, Songxiang Liu, Jiatong Shi, and Tomoki Toda. The singing voice con- version challenge 2023. In 2023 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), pages 1–8. IEEE, 2023

work page 2023