Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

FLowHigh: Towards Efficient and High-Quality Audio Super-Resolution with Single-Step Flow Matching

T0 review · 3 major / 6 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 single Euler step of flow matching outperforms diffusion-based audio super-resolution on VCTK.

desk verdict The formula is right; the real issue is that the SOTA claim rests on a small fixed test split and two baselines evaluated at a different target rate. read the letter →

arxiv 2501.04926 v1 pith:TYNVPHVN submitted 2025-01-09 eess.AS cs.AIcs.CLcs.SD

Jun-Hak Yun , Seung-Bin Kim , Seong-Whan Lee This is my paper · ORCID
classification eess.AScs.AIcs.CLcs.SD
keywords audiosuper-resolutionbandwidthextensionflowmatchingsingle-stepsamplingconditionalmel-spectrogramneuralvocoderdiffusionmodels
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

FLowHigh is an audio super-resolution system that replaces the step-heavy sampling of diffusion models with a single Euler step of conditional flow matching. The paper argues that a flow-matching model operating on mel-spectrograms and conditioned on the low-resolution input reconstructs 48 kHz speech with lower log-spectral distance and higher ViSQOL scores than existing diffusion, GAN, and masked-autoencoder baselines, across input rates from 8 to 24 kHz. If correct, this makes high-quality bandwidth extension practical for low-latency applications, since sampling cost drops from dozens or hundreds of function evaluations to one. The claim is demonstrated on the VCTK speech corpus and relies on a data-dependent prior distribution centered on the input plus a pre-trained neural vocoder for waveform synthesis.

What carries the argument

The engine is conditional flow matching with a specially chosen probability path. The conditional mean interpolates linearly between a source point $x_0$ and a target point $x_1$, with $x_0$ tied to the low-resolution input and $x_1$ the high-resolution target mel-spectrogram, while the standard deviation shrinks as $\sigma_t = 1 - (1-\sigma)t$. This yields the target vector field $u_t = \frac{(x_1-x_0) - (1-\sigma)(x-x_0)}{1-(1-\sigma)t}$. A transformer-based estimator regresses this field at the mel-spectrogram level; at inference, one Euler step moves a sample drawn from $\mathcal{N}(X_h, I)$ along the learned field, and the output is converted to a waveform by a pre-trained BigVGAN vocoder, with an STFT/ISTFT post-processing step that copies the true low-frequency components from the input back into the result.

What would settle it

Compare FLowHigh's output quality at 1, 2, and 10 Euler steps on a held-out set of speakers or on non-speech audio; if quality improves markedly with more steps, the single-step claim relies on a favourable path geometry on VCTK speech rather than a genuinely straight flow.

Watch

Extended reading notes

Core claim

The central discovery is that a conditional flow matching model with a data-dependent prior—a Gaussian source distribution centered on the low-resolution mel-spectrogram—learns a nearly straight vector field between the low-resolution condition and the high-resolution target. Because the path is nearly straight, a single Euler step is enough to land close to the target distribution, yielding state-of-the-art LSD and ViSQOL on VCTK at 48 kHz with one function evaluation. The paper shows this holds for input sampling rates of 8, 12, 16, and 24 kHz, and that the model beats diffusion baselines even when those baselines are given 100 sampling steps.

Load-bearing premise

The result depends on the assumption that a single Euler step along the learned vector field lands close enough to the high-resolution mel-spectrogram distribution that the vocoder and post-processing steps produce high-quality audio.

Editorial extensions

If this is right

  • Flow matching is a practical alternative to diffusion for audio super-resolution, achieving better objective quality with a single function evaluation.
  • Single-step operation reduces the real-time factor of voice bandwidth extension to the cost of one transformer forward pass plus vocoding, enabling low-latency deployment.
  • A data-dependent prior, starting from the low-resolution input, yields better perceptual quality than a standard Gaussian prior in conditional flow matching.
  • The model handles multiple input sampling rates (8, 12, 16, and 24 kHz) with a single estimator, avoiding rate-specific training.
  • Combining a flow-matching estimator with a pre-trained vocoder and low-frequency replacement post-processing is an effective recipe for speech super-resolution.

Reading between the lines

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

  • If the straight-path property is generic, the same single-step recipe could transfer to other ill-posed audio restoration tasks such as speech enhancement or declipping, where a data-dependent prior is available.
  • Because the vector field estimator is a generic transformer, scaling to higher sampling rates or stereo audio may only require a larger vocoder and more training data, not a change in architecture.
  • The paper does not test non-speech audio; a reasonable extension would check whether the data-dependent prior helps or hurts when the target distribution is much broader than speech.
  • The single-step success may be partly due to the VCTK speech domain's narrow bandwidth and speaker distribution; out-of-domain robustness is an open question.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. This paper proposes FLowHigh, an audio super-resolution method based on conditional flow matching operating on mel-spectrograms. The method uses a data-dependent prior centered at the upsampled low-resolution input, a linear interpolation between source and target, and a single-step Euler solver, followed by BigVGAN vocoder synthesis and a low-frequency replacement post-processing. The authors report experiments on VCTK at a 48 kHz target with input rates of 8, 12, 16, and 24 kHz, claiming state-of-the-art LSD and ViSQOL values and large speedups compared to diffusion-based baselines.

Significance. If the empirical claims hold, the paper would be a useful contribution: it demonstrates that a single-step flow-matching model can match or exceed the quality of multi-step diffusion models for speech bandwidth extension, and its analysis of probability paths provides practical guidance for choosing prior distributions in conditional flow matching. The paper also releases code and audio samples, which is commendable. The central derivation of the flow-matching objective is internally consistent; we verified that Eq. (5) follows from Eq. (4). However, the empirical SOTA claim is weakened by the unmatched comparison and the absence of uncertainty estimates.

major comments (3)
  1. [Table I, Section V-A] In Table I, the rows for NVSR and AudioSR are placed under columns labeled '→48 kHz', but the footnote in Section V-A states that these two baselines were evaluated at a 44.1 kHz target rate; consequently, their LSD and ViSQOL numbers are not directly comparable to the 48 kHz results of FLowHigh and the other baselines, and the table should either be restructured to separate the 44.1 kHz results or the baselines should be re-evaluated at 48 kHz.
  2. [Sections IV-A and V] The evaluation is based on a fixed split of only 8 test speakers with a single evaluation run per configuration; no error bars, confidence intervals, or results across random seeds or multiple low-pass filter realizations are reported, so the observed gaps over Fre-painter (e.g., LSD 0.81 vs 0.88 at 8 kHz and 0.71 vs 0.77 at 16 kHz) may fall within run-to-run or speaker-to-speaker variability, which is not established.
  3. [Section V-C and Table III] The probability-path ablation is reported on the same 8-speaker evaluation set used for the final SOTA comparison, and no separate validation set is described; because the choice of probability path is one of the paper's main contributions, selecting it on the test set creates a risk of optimistic bias in the headline results, and the authors should redo the path selection on a held-out validation set and report the final numbers on the test set.
minor comments (6)
  1. [Eq. (5)] Equation (5) is correct, but the algebraic step from Eq. (4) is skipped; adding a short derivation would improve readability.
  2. [Section III-B] The role of x0 is ambiguous: it is called the source point, but it is also sampled from the data-dependent prior p0; the paper should clarify the relationship between the conditioning variable X_h, the prior sample x0, and the flow sample.
  3. [Section V-B] The RTF values in Table II are not accompanied by a description of the evaluation hardware and inference implementation; please specify the GPU and the inference framework.
  4. [Abstract and Conclusion] The abstract and conclusion describe the method as achieving state-of-the-art performance without mentioning the 44.1 kHz caveat for two baselines; the claims should be moderated to reflect the matched-rate comparison.
  5. [Implementation Details] The paper gives the model parameter count but does not report the total training wall-clock time; please provide this information.
  6. [Table I caption] The caption contains a LaTeX artifact ('T HE'); please fix the caption formatting.

Circularity Check

1 steps flagged · score 3.0 of 10

No circularity in the flow-matching derivation; the only circular step is the probability-path choice being selected on the same VCTK evaluation set later used for the SOTA claim.

  1. fitted input called prediction [Section V-C, Table III and Section V-A, Table I; Section IV-A defines the evaluation set.]
    "We performed audio SR using various probability paths to validate the proposed path, with the evaluation results presented in Table III."

    The proposed probability path is chosen by comparing LSD and ViSQOL on the same 8-speaker VCTK evaluation set that is later used to report the headline state-of-the-art numbers in Table I. The manuscript describes no separate validation split, so the selection of the path is effectively fit to the evaluation set's metrics. The final SOTA claim is therefore not an out-of-sample prediction of a pre-specified model; it is a report on the data that was used to select the model variant. This is a mild form of fitted-input-called-prediction: the design choice of mu_t and sigma_t is tuned to the test set, then the test set is presented as the evidence of superiority.

full rationale

The paper's core derivation is not circular. With mu_t = t*x1 + (1-t)*x0 and sigma_t = 1-(1-sigma)*t, Eq. (4) gives u_t = ((x1-x0) - (1-sigma)*(x-x0)) / (1-(1-sigma)*t), which is exactly Eq. (5); the training objective Eq. (6) regresses the estimator onto this target vector field, and the single-step Euler inference follows the learned flow. Comparisons to baselines use official implementations and pre-trained checkpoints on the public VCTK corpus, so the central SOTA claim is externally checkable rather than forced by self-citation. The only circularity-adjacent issue is that the choice among probability paths in Table III is made using the same evaluation set that later produces the headline metrics in Table I, with no separate validation split described; this leaks test information into model selection and makes the final numerical claim less independent than it appears. Because the model must still be trained and the result is not forced by construction, the circularity is mild.

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

The central claim depends on standard flow matching theory, a pre-trained vocoder, and several hand-chosen hyperparameters. No new physical or conceptual entities are introduced. The largest unstated dependency is the exact post-processing rule, which is load-bearing for the low-frequency metrics.

free parameters (5)
  • sigma = 1e-4
    Noise floor in the conditional path of Eq. (5), hand-chosen as sufficiently small for all experiments.
  • training LR sampling-rate range = 4-32 kHz random
    Data augmentation range for the input sampling rate, chosen by hand and affecting generalization across evaluation rates.
  • STFT parameters = window 2048, hop 480, FFT 2048
    Fixed analysis parameters that define the mel-spectrogram representation used by the model.
  • model hyperparameters = 2 layers, 16 heads, 1024 embedding dim, 4096 FFN
    Architecture choices for the transformer vector field estimator; described without tuning analysis.
  • post-processing crossover rule = not specified
    The exact rule for replacing low-frequency components from the input is not stated, yet it materially affects LSD-LF results.
assumptions (5)
  • standard math Flow matching theory: the conditional flow matching objective has the same gradients as the flow matching objective.
    Invoked in Section II from Lipman et al. [29] and Tong et al. [40].
  • domain assumption VCTK train/test speaker split is representative and free of speaker leakage.
    Section IV-A: 100 speakers for training, 8 for evaluation, with no overlap stated.
  • domain assumption Mel-spectrogram plus the pre-trained BigVGAN vocoder is a sufficient representation for high-fidelity 48 kHz audio.
    Section III-A and III-D rely on the vocoder from [44] and the two-stage pipeline from [26].
  • domain assumption STFT/ISTFT replacement of low-frequency components fully preserves the original LR band.
    Section III-D applies post-processing without an error analysis of the replacement step.
  • domain assumption LSD and ViSQOL adequately capture reconstruction and perceptual quality for audio super-resolution.
    Section IV-C defines the evaluation metrics without comparing them to human listening.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FLowHigh: Towards Efficient and High-Quality Audio Super-Resolution with Single-Step Flow Matching." pith.science (2026). https://pith.science/paper/TYNVPHVN

@misc{pith2026250104926,
  author       = {Pith},
  title        = {Pith review of: FLowHigh: Towards Efficient and High-Quality Audio Super-Resolution with Single-Step Flow Matching},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TYNVPHVN}},
  note         = {Machine review of arXiv:2501.04926}
}
read the original abstract

Audio super-resolution is challenging owing to its ill-posed nature. Recently, the application of diffusion models in audio super-resolution has shown promising results in alleviating this challenge. However, diffusion-based models have limitations, primarily the necessity for numerous sampling steps, which causes significantly increased latency when synthesizing high-quality audio samples. In this paper, we propose FLowHigh, a novel approach that integrates flow matching, a highly efficient generative model, into audio super-resolution. We also explore probability paths specially tailored for audio super-resolution, which effectively capture high-resolution audio distributions, thereby enhancing reconstruction quality. The proposed method generates high-fidelity, high-resolution audio through a single-step sampling process across various input sampling rates. The experimental results on the VCTK benchmark dataset demonstrate that FLowHigh achieves state-of-the-art performance in audio super-resolution, as evaluated by log-spectral distance and ViSQOL while maintaining computational efficiency with only a single-step sampling process.

Figures

Figures reproduced from arXiv: 2501.04926 by the authors.

Figure 1
Figure 1. Overview of FLowHigh. (a) The overall training and inference process of FLowHigh based on conditional flow matching. (b) The ordinary differential [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Spectrogram visualizations of ground truth, input signal, outputs from [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. A2SB: Audio-to-Audio Schrodinger Bridges

    cs.SD 2025-01 conditional novelty 6.0 of 10

    A2SB applies Schrödinger bridges to music restoration, achieving state-of-the-art bandwidth extension and inpainting at 44.1kHz in a single vocoder-free model.

Reference graph

Works this paper leans on

46 extracted references · 37 canonical work pages · cited by 1 Pith paper

  1. [1]

    Waveform modeling and generation using hierarchical recurrent neural networks for speech bandwidth extension,

    Zhen-Hua Ling, Yang Ai, Yu Gu, and Li-Rong Dai, “Waveform modeling and generation using hierarchical recurrent neural networks for speech bandwidth extension,” IEEE/ACM Trans. Audio, Speech, Lang. Process., vol. 26, pp. 883–894, 2018

    work page 2018

  2. [2]

    A simple cepstral domain dnn approach to artificial speech bandwidth extension,

    Johannes Abel, Maximilian Strake, and Tim Fingscheidt, “A simple cepstral domain dnn approach to artificial speech bandwidth extension,” in IEEE Int. Conf. Acoust. Speech Signal Process. (ICASSP) , 2018

    work page 2018

  3. [3]

    Temporal film: Capturing long-range sequence dependencies with feature-wise modulations,

    Sawyer Birnbaum, V olodymyr Kuleshov, Zayd Enam, Pang Wei W Koh, and Stefano Ermon, “Temporal film: Capturing long-range sequence dependencies with feature-wise modulations,” in Adv. Neural Inf. Process. Syst. (NeurIPS) , 2019, vol. 32

    work page 2019

  4. [4]

    Towards robust speech super- resolution,

    Heming Wang and DeLiang Wang, “Towards robust speech super- resolution,” IEEE/ACM Trans. Audio, Speech, Lang. Process. , vol. 29, pp. 2058–2066, 2021

    work page 2021

  5. [5]

    Multiresolution recognition of unconstrained handwritten numerals with wavelet transform and multilayer cluster neural network,

    Seong-Whan Lee, Chang-Hun Kim, Hong Ma, and Yuan Y Tang, “Multiresolution recognition of unconstrained handwritten numerals with wavelet transform and multilayer cluster neural network,” Pattern Recognit., vol. 29, pp. 1953–1961, 1996

    work page 1953

  6. [6]

    Classification of drowsiness levels based on a deep spatio-temporal convolutional bidirectional LSTM network using electroencephalogra- phy signals,

    Ji-Hoon Jeong, Baek-Woon Yu, Dae-Hyeok Lee, and Seong-Whan Lee, “Classification of drowsiness levels based on a deep spatio-temporal convolutional bidirectional LSTM network using electroencephalogra- phy signals,” Brain Sci., vol. 9, pp. 348, 2019

    2019

  7. [7]

    A multi-view cnn with novel variance layer for motor imagery brain computer interface,

    Ravikiran Mane, Neethu Robinson, A Prasad Vinod, Seong-Whan Lee, and Cuntai Guan, “A multi-view cnn with novel variance layer for motor imagery brain computer interface,” in Proc. Int. Conf. IEEE Eng. Med. Biol. Soc. (EMBC) , 2020

    2020

  8. [8]

    WSRGlow: A glow-based waveform generative model for audio super-resolution,

    Kexun Zhang, Yi Ren, Changliang Xu, and Zhou Zhao, “WSRGlow: A glow-based waveform generative model for audio super-resolution,” in Ann. Conf. Int. Speech Commun. Assoc. (INTERSPEECH) , 2021

    work page 2021

Show all 46 references
  1. [9]

    NU-Wave: A diffusion probabilistic model for neural audio upsampling,

    Junhyeok Lee and Seungu Han, “NU-Wave: A diffusion probabilistic model for neural audio upsampling,” in Ann. Conf. Int. Speech Commun. Assoc. (INTERSPEECH), 2021

    2021

  2. [10]

    V oiceFixer: A unified framework for high-fidelity speech restoration,

    Haohe Liu et al., “V oiceFixer: A unified framework for high-fidelity speech restoration,” in Ann. Conf. Int. Speech Commun. Assoc. (INTER- SPEECH), 2022

    2022

  3. [11]

    AERO: Audio super resolution in the spectral somain,

    Moshe Mandel, Or Tal, and Yossi Adi, “AERO: Audio super resolution in the spectral somain,” in IEEE Int. Conf. Acoust. Speech Signal Process. (ICASSP), 2023

    2023

  4. [12]

    mdct- GAN: Taming transformer-based GAN for speech super-resolution with Modified DCT spectra,

    Chenhao Shuai, Chaohua Shi, Lu Gan, and Hongqing Liu, “mdct- GAN: Taming transformer-based GAN for speech super-resolution with Modified DCT spectra,” in Ann. Conf. Int. Speech Commun. Assoc. (INTERSPEECH), 2023

    2023

  5. [13]

    Multi- stage speech bandwidth extension with flexible sampling rate control,

    Ye-Xin Lu, Yang Ai, Zheng-Yan Sheng, and Zhen-Hua Ling, “Multi- stage speech bandwidth extension with flexible sampling rate control,” in Ann. Conf. Int. Speech Commun. Assoc. (INTERSPEECH) , 2024

    2024

  6. [14]

    Audio super-resolution with robust speech representation learning of masked autoencoder,

    Seung-Bin Kim, Sang-Hoon Lee, Ha-Yeong Choi, and Seong-Whan Lee, “Audio super-resolution with robust speech representation learning of masked autoencoder,” IEEE/ACM Trans. Audio, Speech, Lang. Process., vol. 32, pp. 1012–1022, 2024

    2024

  7. [15]

    Denoising diffusion probabilistic models,

    Jonathan Ho, Ajay Jain, and Pieter Abbeel, “Denoising diffusion probabilistic models,” in Adv. Neural Inf. Process. Syst. (NeurIPS), 2020

    2020

  8. [16]

    Denoising diffusion implicit models,

    Jiaming Song, Chenlin Meng, and Stefano Ermon, “Denoising diffusion implicit models,” in Int. Conf. Learn. Represent. (ICLR) , 2020

    2020

  9. [17]

    Score-based generative modeling through stochastic differential equations,

    Yang Song et al., “Score-based generative modeling through stochastic differential equations,” in Int. Conf. Learn. Represent. (ICLR) , 2021

    2021

  10. [18]

    NU-Wave 2: A general neural audio upsampling model for various sampling rates,

    Seungu Han and Junhyeok Lee, “NU-Wave 2: A general neural audio upsampling model for various sampling rates,” in Ann. Conf. Int. Speech Commun. Assoc. (INTERSPEECH) , 2022

    2022

  11. [19]

    Condi- tioning and sampling in variational diffusion models for speech super- resolution,

    Chin-Yun Yu, Sung-Lin Yeh, Gy ¨orgy Fazekas, and Hao Tang, “Condi- tioning and sampling in variational diffusion models for speech super- resolution,” in IEEE Int. Conf. Acoust. Speech Signal Process. (ICASSP), 2023

    2023

  12. [20]

    AudioSR: Versatile audio super-resolution at scale,

    Haohe Liu, Ke Chen, Qiao Tian, Wenwu Wang, and Mark D Plumbley, “AudioSR: Versatile audio super-resolution at scale,” in IEEE Int. Conf. Acoust. Speech Signal Process. (ICASSP) , 2024

    2024

  13. [21]

    Grad-TTS: A diffusion probabilistic model for text- to-speech,

    Vadim Popov, Ivan V ovk, Vladimir Gogoryan, Tasnima Sadekova, and Mikhail Kudinov, “Grad-TTS: A diffusion probabilistic model for text- to-speech,” in Proc. Int. Conf. Mach. Learn. (ICML) , 2021

    2021

  14. [22]

    AudioLDM: Text-to-audio generation with latent diffusion models,

    Haohe Liu et al., “AudioLDM: Text-to-audio generation with latent diffusion models,” in Proc. Int. Conf. Mach. Learn. (ICML) , 2023

    2023

  15. [23]

    AudioLDM 2: Learning holistic audio generation with self-supervised pretraining,

    Haohe Liu et al., “AudioLDM 2: Learning holistic audio generation with self-supervised pretraining,” IEEE/ACM Transactions on Audio, Speech, and Language Processing , vol. 32, pp. 2871–2883, 2024

    2024

  16. [24]

    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 IEEE Int. Conf. Acoust. Speech Signal Process. (ICASSP), 2024

    2024

  17. [25]

    Diffprosody: Diffusion-based latent prosody generation for expressive speech synthe- sis with prosody conditional adversarial training,

    Hyung-Seok Oh, Sang-Hoon Lee, and Seong-Whan Lee, “Diffprosody: Diffusion-based latent prosody generation for expressive speech synthe- sis with prosody conditional adversarial training,” IEEE/ACM Trans. Audio, Speech, Lang. Process. , vol. 32, pp. 2654–2666, 2024

    2024

  18. [26]

    Neural vocoder is all you need for speech super- resolution,

    Haohe Liu et al., “Neural vocoder is all you need for speech super- resolution,” in Ann. Conf. Int. Speech Commun. Assoc. (INTER- SPEECH), 2022

    2022

  19. [27]

    Towards an EEG-based intuitive BCI communication system using imagined speech and visual imagery,

    Seo-Hyun Lee, Minji Lee, Ji-Hoon Jeong, and Seong-Whan Lee, “Towards an EEG-based intuitive BCI communication system using imagined speech and visual imagery,” in Proc. IEEE Int. Conf. Syst. Man Cybern. (SMC) , 2019

    2019

  20. [28]

    An adaptive deep reinforcement learning framework enables curling robots with human-like performance in real-world conditions,

    Dong-Ok Won, Klaus-Robert M ¨uller, and Seong-Whan Lee, “An adaptive deep reinforcement learning framework enables curling robots with human-like performance in real-world conditions,” Sci. Robot., vol. 5, pp. eabb9764, 2020

    2020

  21. [29]

    Flow matching for generative modeling,

    Yaron Lipman, Ricky T. Q. Chen, Heli Ben-Hamu, Maximilian Nickel, and Matthew Le, “Flow matching for generative modeling,” in Int. Conf. Learn. Represent. (ICLR) , 2023

    2023

  22. [30]

    Flow straight and fast: learning to generate and transfer data with rectified flow,

    Xingchao Liu, Chengyue Gong, and qiang liu, “Flow straight and fast: learning to generate and transfer data with rectified flow,” in Int. Conf. Learn. Represent. (ICLR) , 2023

    2023

  23. [31]

    Neural ordinary differential equations,

    Ricky T. Q. Chen, Yulia Rubanova, Jesse Bettencourt, and David K Duvenaud, “Neural ordinary differential equations,” in Adv. Neural Inf. Process. Syst. (NeurIPS) , 2018

    2018

  24. [32]

    V oicebox: Text-guided multilingual universal speech generation at scale,

    Matthew Le et al., “V oicebox: Text-guided multilingual universal speech generation at scale,” in Adv. Neural Inf. Process. Syst. (NeurIPS) , 2023

    2023

  25. [33]

    Audiobox: Unified audio generation with natural language prompts,

    Apoorv Vyas et al., “Audiobox: Unified audio generation with natural language prompts,” arXiv preprint arXiv:2312.15821 , 2023

    2023 arXiv

  26. [34]

    Generative pre-training for speech with flow matching,

    Alexander H. Liu et al., “Generative pre-training for speech with flow matching,” in Int. Conf. Learn. Represent. (ICLR) , 2024

    2024

  27. [35]

    Matcha-TTS: A fast TTS architecture with conditional flow matching,

    Shivam Mehta, Ruibo Tu, Jonas Beskow, ´Eva Sz ´ekely, and Gustav Eje Henter, “Matcha-TTS: A fast TTS architecture with conditional flow matching,” in IEEE Int. Conf. Acoust. Speech Signal Process. (ICASSP) , 2024

    2024

  28. [36]

    MusicFlow: Cascaded flow matching for text guided music generation,

    K. R. Prajwal et al., “MusicFlow: Cascaded flow matching for text guided music generation,” in Proc. Int. Conf. Mach. Learn. (ICML) , 2024

    2024

  29. [37]

    PeriodWave: Multi-period flow matching for high-fidelity waveform generation,

    Sang-Hoon Lee, Ha-Yeong Choi, and Seong-Whan Lee, “PeriodWave: Multi-period flow matching for high-fidelity waveform generation,” arXiv preprint arXiv:2408.07547 , 2024

    2024 arXiv

  30. [38]

    EmoSphere++: Emotion-controllable zero-shot text-to-speech via emotion-adaptive spherical vector,

    Deok-Hyeon Cho, Hyung-Seok Oh, Seung-Bin Kim, and Seong-Whan Lee, “EmoSphere++: Emotion-controllable zero-shot text-to-speech via emotion-adaptive spherical vector,” arXiv preprint arXiv:2411.02625 , 2024

    2024 arXiv

  31. [39]

    FlowA VSE: Efficient audio-visual speech enhancement with conditional flow matching,

    Chaeyoung Jung, Suyeon Lee, Ji-Hoon Kim, and Joon Son Chung, “FlowA VSE: Efficient audio-visual speech enhancement with conditional flow matching,” in Ann. Conf. Int. Speech Commun. Assoc. (INTER- SPEECH), 2024

    2024

  32. [40]

    Improving and generalizing flow-based gener- ative models with minibatch optimal transport,

    Alexander Tong et al., “Improving and generalizing flow-based gener- ative models with minibatch optimal transport,” Trans. Mach. Learn. Res., 2024

    2024

  33. [41]

    Multisample flow matching: Straight- ening flows with minibatch couplings,

    Aram-Alexandre Pooladian et al., “Multisample flow matching: Straight- ening flows with minibatch couplings,” in Proc. Int. Conf. Mach. Learn. (ICML), 2023

    2023

  34. [42]

    Attention is all you need,

    Ashish Vaswani et al., “Attention is all you need,” in Adv. Neural Inf. Process. Syst. (NeurIPS) , 2017

    2017

  35. [43]

    CSTR VCTK Corpus: English multi-speaker corpus for CSTR voice cloning toolkit,

    Junichi Yamagishi, Christophe Veaux, and Kirsten MacDonald, “CSTR VCTK Corpus: English multi-speaker corpus for CSTR voice cloning toolkit,” 2019, The Centre for Speech Technology Research (CSTR), Univ. Edinburgh, doi: 10.7488/ds/2645

    2019 doi

  36. [44]

    BigVGAN: A universal neural vocoder with large-scale training,

    Sang gil Lee, Wei Ping, Boris Ginsburg, Bryan Catanzaro, and Sungroh Yoon, “BigVGAN: A universal neural vocoder with large-scale training,” in Int. Conf. Learn. Represent. (ICLR) , 2023

    2023

  37. [45]

    Adam: A method for stochastic optimization,

    Diederik Kingma and Jimmy Ba, “Adam: A method for stochastic optimization,” in Int. Conf. Learn. Represent. (ICLR) , 2015

    2015

  38. [46]

    ViSQOL v3: An open source production ready objective speech and audio metric,

    Michael Chinen et al., “ViSQOL v3: An open source production ready objective speech and audio metric,” in IEEE Int. Conf. Qual. Multimedia Experience (QoMEX), 2020

    2020

Pith tools

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