Pith. sign in

REVIEW 3 major objections 5 minor 34 references

Speech Synthesis along Perceptual Voice Quality Dimensions

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

Pith's one-line read A text-to-speech system can turn perceptual voice qualities — breathiness, roughness, resonance, and weight — up or down on a continuous scale by conditioning a normalizing flow on perceptual labels.

desk verdict A solid, genuinely new voice-manipulation system whose breathiness and roughness claims are validated by experts, but whose resonance and weight claims rest on a circular metric. read the letter →

arxiv 2501.08791 v1 pith:JZOIVTA2 submitted 2025-01-15 eess.AS cs.SD

classification eess.AScs.SD
keywords perceptualvoicequalitiestext-to-speechconditionalcontinuousnormalizingflowsmodificationbreathinessroughnessresonancespeakerembeddings
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 aims to show that a text-to-speech system can manipulate perceptual voice qualities — breathiness, roughness, resonance, and weight — along a continuous severity scale, without directly editing acoustic correlates such as jitter or shimmer. The authors insert a conditional continuous normalizing flow between the speaker encoder and the decoder of an existing multi-speaker TTS model, and train the flow to move a speaker embedding according to an attribute vector. The attribute vectors come from statistical predictor models that pseudo-label a large read-speech corpus, so the model learns the manipulation from examples rather than from hand-designed acoustic rules. Phonetic experts rated the synthesized voices and found breathiness shifts convincing, while roughness required stronger settings; speaker similarity degrades as severity increases. If the approach holds, it gives clinicians and voice practitioners a label-driven way to generate graded voice-quality examples for training and data augmentation.

What carries the argument

The load-bearing object is the Conditional Continuous Normalizing Flow (CCNF), a generative model that turns samples into a standard normal distribution by solving an ordinary differential equation, and can generate new samples by solving the ODE backwards under a changed conditioning vector. Here the conditioning vector has eight entries: seven perceptual voice qualities plus speaking rate. The flow is inserted between a fixed speaker-embedding extractor and the TTS decoder: it first encodes the original embedding under the original attribute vector, then decodes it under the manipulated vector, so the ODE trajectory itself carries the requested voice-quality change. This design is what lets the system manipulate qualities without ever touching acoustic correlates directly.

What would settle it

Take a held-out set of ordinary read-speech utterances, have three professional voice-rating experts score breathiness and roughness on the 100-point clinical scale, and compare the ordering of their scores with the ordering of the pseudo-labels used in training. If the orderings disagree substantially, the conditioning values do not encode the claimed perceptual qualities, and the manipulation results would not generalize beyond the pseudo-label distribution.

Watch

Extended reading notes

Core claim

The central claim is that a Conditional Continuous Normalizing Flow, trained to map speaker embeddings under an attribute vector, can act as a voice-quality dial inside a TTS system: at synthesis time the original speaker embedding is transformed to a latent Gaussian with the original attributes and then transformed back with the desired attributes, and the difference between those two paths is the modification. The paper demonstrates this on four perceptual voice qualities and reports that the manipulated severity, as predicted by the regressors, rises monotonically with the requested level, and that the shift is statistically significant for breathiness at all levels and for roughness only at strong modifications. The work also claims to generalize to unseen speakers, because the modifications operate on the speaker embedding rather than on the acoustic waveform, and to preserve audio quality except under extreme settings. Expert-rated breathiness reached roughly 70–79 on the 100-point clinical scale from a typical-voice baseline near 24, which the authors take as evidence that the label-driven manipulation is perceptually real.

Load-bearing premise

The labels used to train the voice-quality dial come from statistical models trained on only 296 atypical-voice recordings; if those models misjudge ordinary read speech, the dial is tuned to qualities that do not match what listeners hear.

Editorial extensions

If this is right

  • Clinicians could generate graded breathy or rough voice examples for training speech pathologists, using ordinary read speech as the starting point.
  • Voice-quality editing can work for speakers never seen during training, because the manipulation lives in speaker-embedding space rather than in a per-speaker model.
  • Strong manipulations move the voice away from the original speaker, so the method can also prod how much of speaker identity is carried by voice quality.
  • The usable range differs per quality: breathiness moves convincingly at moderate settings, while roughness needs aggressive conditioning, so a deployment tool would need per-quality calibration.

Reading between the lines

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

  • A natural extension is that adding a new perceptual quality is mostly a matter of obtaining reliable predictor models and pseudo-labels; the flow-training pipeline itself need not be redesigned.
  • The observed covariation between roughness and breathiness could be exploited to study how perceptual voice qualities interact, since the flow lets one axis be moved while others are held fixed.
  • The same design pattern suggests a broader recipe: any attribute that can be pseudo-labeled on a large unlabeled speech corpus could become a controllable TTS dimension, as long as it leaves a trace in the speaker embedding.
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 integrates a Conditional Continuous Normalizing Flow (CCNF) into YourTTS to manipulate perceptual voice qualities (PVQs) on a continuous scale, targeting roughness, breathiness, resonance, and weight. Speaker embeddings are transformed by conditioning on an attribute vector of PVQ severities (plus speaking rate), which is estimated by random forest regressors trained on the PVQD+ dataset and used to pseudo-label LibriTTS-R. The method is evaluated objectively with the same regressors and with EER speaker-similarity measures, and subjectively by phonetic experts for roughness and breathiness only. The authors report significant perceived changes in breathiness across all levels, significant roughness change only at the highest extrapolated setting, modest MOS degradation, and increasing EER with strong manipulation.

Significance. If the result holds, the system would be a useful tool for generating graded voice-quality examples for training speech pathologists and for data augmentation in atypical-voice processing, and it would extend TTS controllability beyond prosody and emotion to lower-level perceptual attributes without hand-crafted acoustic manipulations. The work has clear strengths: the CCNF approach avoids direct acoustic-correlate manipulation; the system is evaluated on both seen and unseen speakers; and the subjective expert listening test provides genuinely independent evidence for two of the four qualities. The main weakness is that the evidence for the other two qualities (resonance and weight) rests entirely on a circular objective metric, so the paper's broad conclusion that all four PVQs are 'effectively manipulated' is not yet supported.

major comments (3)
  1. [Section III-B, Table II] The objective evaluation of resonance and weight is circular: the random forest regressors trained on PVQD+ are used both to generate the pseudo-labels for LibriTTS-R training data and to evaluate the manipulated speech in Table II. For resonance and weight, Table II is the only reported evidence, so the predicted severities largely measure whether the CCNF can invert the regressors' own mapping rather than whether listeners perceive the intended changes. The paper itself notes the compression of predicted scores relative to expert ratings (e.g., breathiness 11/19/35 vs. 24/46/70 in Tables II and III). Since no expert validation is provided for resonance or weight, the Section V conclusion that the system 'effectively manipulates PVQs' overreaches for these two qualities. Please provide independent perceptual evidence for resonance and weight, or explicitly restrict the perceptual claim to breathiness and roughness.
  2. [Section IV-A, Table III] The roughness manipulation result does not support a continuous-control claim. Only the 'high' condition, which uses an extrapolated scale value of 200 beyond the CAPE-V maximum of 100, produces a statistically significant perceived change (p < 0.005); low and medium conditions are not significantly different from the original recording. The paper should either present the roughness claim as limited to strong extrapolated settings or provide additional evidence that intermediate levels are perceptually distinguishable. The choice of 0/100/200 is also an ad-hoc modification of the evaluation scale that deserves explicit justification.
  3. [Section III-B] The pseudo-labeling pipeline is a load-bearing risk. Random forest regressors are trained on only 296 PVQD+ files (containing atypical voices) and then applied to 585 hours of typical LibriTTS-R speech. If the regressors' HuBERT-based features do not generalize from atypical to typical voices, the pseudo-labels themselves are misaligned with the perceptual qualities the system claims to control. The fact that predicted severities for original typical speech are strongly compressed relative to expert ratings (e.g., breathiness 11 vs. expert 24) suggests this risk is real. For breathiness and roughness, the expert listening test partly mitigates the concern, but for resonance and weight no external check exists. Please report the distribution of pseudo-labels on LibriTTS-R and, if possible, a validation of the regressors on typical-voice samples with expert ratings.
minor comments (5)
  1. [Section IV-A] The text refers to a 'paired Wilcoxon rank sum test'; this is a terminological inconsistency, as 'rank sum' (Mann-Whitney U) is for unpaired data while a paired test would be the Wilcoxon signed-rank test. Please correct the wording.
  2. [Table II] The EER values for resonance and weight are strongly non-monotonic (e.g., seen resonance: 8.8, 1.4, 22.2; seen weight: 29.6, 1.7, 19.4). The explanation in Section IV-C about gender spoofing is plausible, but the table would benefit from confidence intervals or a statistical comparison to establish that the medium-condition drop is meaningful and not noise.
  3. [Section II, Eq. (3)] The definition of Δlogp(θ) appears to have a sign inconsistency: the initial condition sets log pZ1(z(t1)) - log pS(s|a; θ) = 0, yet this term is added to the trace integral. Please check the signs and clarify the relation between Δlogp and the standard instantaneous change-of-variables formula.
  4. [Section IV] The listening-test design description is not fully consistent: 15 raters x 32 samples = 480 total, and with 4 speakers and conditions (roughness 3, breathiness 3, plus original) the per-speaker count appears to be 7, not 8. Please clarify the exact number of conditions per speaker.
  5. [Section III-B] Reference [29] is cited for the choice of the 6th HuBERT layer, but [29] is the WavLM paper; please ensure the citation correctly supports the HuBERT feature claim, and align the text with the cited source.

Circularity Check

1 steps flagged · score 6.0 of 10

Objective PVQ evaluation for resonance and weight reuses the same regressors that generated the training pseudo-labels, making those results circular; expert ratings only validate two of the four qualities.

  1. fitted input called prediction [Section III-B (pseudo-labeling) and Section IV-A (objective evaluation), Tables II-III]
    "These fitted regressors were used to pseudo-label the LibriTTS-R dataset. ... Objective evaluation included Equal Error Rate (EER) to measure speaker similarity and the random forest regressors used for pseudo-labeling to evaluate the severity of the modified PVQs."

    The CCNF is trained to map speaker embeddings conditioned on regressor-generated pseudo-labels, and the same random forest regressors are then used as the objective severity measure after synthesis. For resonance and weight, which have no expert ratings, Table II therefore only shows that the system can move audio such that the original label-generating regressor outputs higher scores; it does not show that listeners perceive the intended PVQ. The monotonic increase is an internal-consistency check against the training label generator, not an independent prediction. The paper's own caveat about the regressor score skew reinforces that these scores are not ground truth. Only the roughness and breathiness expert ratings in Table III break this loop, and only for those two qualities.

full rationale

The core circularity is the reuse of the PVQD+ random forest regressors in both training-label generation and objective evaluation. Section III-B states that the regressors pseudo-label LibriTTS-R, and Section IV states that the same regressors evaluate the modified PVQs. Thus the 'predicted PVQ' numbers in Table II are not independent evidence for any PVQ; they measure whether the synthesized audio moves along the regressor-defined axis used to create the training targets. This is partial rather than total circularity because the subjective listening test with phonetic experts independently confirms breathiness and roughness manipulation (Table III), and the method itself is not defined in terms of the evaluation outcome. However, the conclusion claims success for all four PVQs, while resonance and weight rest only on the circular objective loop. No load-bearing self-citation was found: the only author-overlapping reference is the d-vector speaker encoder, which is an auxiliary component and not the basis of the PVQ-manipulation claim. The paper's own limitation statement about regressor score skew ('demanding awareness when relying on regression models to judge voice quality') further supports treating the objective scores with caution.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The central claim rests on an unvalidated pseudo-labeling chain, one ad hoc extrapolation of the roughness scale, and the standard mathematical apparatus of normalizing flows. No new physical entities are introduced. The free parameters are the flow's hidden size and the adjusted roughness evaluation scale.

free parameters (2)
  • Roughness manipulation scale = 0, 100, 200
    Ad hoc adjustment in Section IV because the model could not produce high perceived roughness at the standard severity of 100, attributed to under-representation of extreme roughness in training data. This extrapolates beyond the CAPE-V 0-100 scale and beyond the training label range.
  • CCNF hidden size = 512
    The hidden size of the flow network was chosen by monitoring overfitting rather than by a principled method; it affects model capacity but not the scientific claim directly.
assumptions (5)
  • domain assumption Speaker embeddings are samples from pS(sn|xn, an) and xn is implicitly conditioned on an, so the attribute vector need not be supplied separately to the speaker encoder.
    Section III-B states this assumption directly. If PVQ information is not sufficiently encoded in the d-vector/HuBERT-derived speaker embedding, the flow has no way to learn a mapping from attributes to identity codes.
  • domain assumption Random forest regressors trained on PVQD+ generalize to LibriTTS-R, so pseudo-labels on typical read speech are accurate enough to serve as training targets.
    Section III-B describes training regressors on PVQD+ and applying them to LibriTTS-R. The regressors have test RMSE around 11.28, and the distribution of non-gendered PVQ values in LibriTTS-R is skewed toward low severity, so this generalization is fragile.
  • standard math Continuous normalizing flows with ODE solvers and Hutchinson trace estimation provide valid density modeling and attribute conditioning.
    Section II reviews CCNFs and relies on change-of-variables, neural ODE solvers, and trace estimation from [18], [20]-[22].
  • ad hoc to paper The CCNF can extrapolate beyond its training distribution to realize roughness values above the CAPE-V maximum.
    Section IV adjusts the roughness modification range to 0, 100, 200 because the model struggled to map the full strength spectrum. This assumes the flow produces meaningful outputs far outside observed labels.
  • domain assumption CAPE-V 100-point severity ratings and the PVQD+ annotations used to train the regressors are meaningful ground truth for the four PVQs.
    The whole pseudo-labeling chain starts from these ratings; the paper acknowledges high annotation variability in [15].

how reviews work

0 comments
Cite this review

Pith. "Pith review of Speech Synthesis along Perceptual Voice Quality Dimensions." pith.science (2026). https://pith.science/paper/JZOIVTA2

@misc{pith2026250108791,
  author       = {Pith},
  title        = {Pith review of: Speech Synthesis along Perceptual Voice Quality Dimensions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JZOIVTA2}},
  note         = {Machine review of arXiv:2501.08791}
}
read the original abstract

While expressive speech synthesis or voice conversion systems mainly focus on controlling or manipulating abstract prosodic characteristics of speech, such as emotion or accent, we here address the control of perceptual voice qualities (PVQs) recognized by phonetic experts, which are speech properties at a lower level of abstraction. The ability to manipulate PVQs can be a valuable tool for teaching speech pathologists in training or voice actors. In this paper, we integrate a Conditional Continuous-Normalizing-Flow-based method into a Text-to-Speech system to modify perceptual voice attributes on a continuous scale. Unlike previous approaches, our system avoids direct manipulation of acoustic correlates and instead learns from examples. We demonstrate the system's capability by manipulating four voice qualities: Roughness, breathiness, resonance and weight. Phonetic experts evaluated these modifications, both for seen and unseen speaker conditions. The results highlight both the system's strengths and areas for improvement.

Figures

Figures reproduced from arXiv: 2501.08791 by the authors.

Figure 1
Figure 1. TTS system during inference [19]. A speaker manipulation block is [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Speaker embedding manipulation block, where [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 33 canonical work pages

  1. [1]

    Wavebender GAN: An Architecture for Phonetically Meaningful Speech Manipulation,

    G. Teodoro D ¨ohler Beck, U. Wennberg, Z. Malisz, and G. Eje Hen- ter, “Wavebender GAN: An Architecture for Phonetically Meaningful Speech Manipulation,” in ICASSP 2022-2022 IEEE International Con- ference on Acoustics, Speech and Signal Processing (ICASSP) , 2022, pp. 6187–6191

  2. [2]

    Hierarchical Intonation Modelling for Speech Synthesis using Legendre Polynomial Coefficients,

    J. O’Mahony, N. Corkey, C. Lai, E. Klabbers, and S. King, “Hierarchical Intonation Modelling for Speech Synthesis using Legendre Polynomial Coefficients,” in Proc. Speech Prosody 2024 , 2024, pp. 1030–1034

  3. [3]

    Prosody-Controllable Spontaneous TTS with Neural HMMs,

    H. Lameris, S. Mehta, G. E. Henter, J. Gustafson, and ´E. Sz ´ekely, “Prosody-Controllable Spontaneous TTS with Neural HMMs,” in ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , 2023, pp. 1–5

  4. [4]

    Towards End-to-End Prosody Transfer for Expressive Speech Synthesis with Tacotron,

    R. Skerry-Ryan, E. Battenberg, Y . Xiao, Y . Wang, D. Stanton, J. Shor, R. Weiss, R. Clark, and R. A. Saurous, “Towards End-to-End Prosody Transfer for Expressive Speech Synthesis with Tacotron,” in Interna- tional Conference on Machine Learning . PMLR, 2018, pp. 4693–4702

  5. [5]

    An Overview of Affective Speech Synthesis and Conversion in the Deep Learning Era,

    A. Triantafyllopoulos, B. W. Schuller, G. ˙Iymen, M. Sezgin, X. He, Z. Yang, P. Tzirakis, S. Liu, S. Mertes, E. Andr ´e, et al., “An Overview of Affective Speech Synthesis and Conversion in the Deep Learning Era,” Proceedings of the IEEE , vol. 111, no. 10, pp. 1355–1381, 2023

  6. [6]

    Daisy-TTS: Simulating Wider Spectrum of Emotions via Prosody Embedding Decomposition

    R. Chevi and A. F. Aji, “Daisy-TTS: Simulating Wider Spectrum of Emotions via Prosody Embedding Decomposition,” arXiv preprint arXiv:2402.14523, 2024

  7. [7]

    Singh, Profiling Humans from their Voice , vol

    R. Singh, Profiling Humans from their Voice , vol. 41, Springer, 2019

  8. [8]

    The Analysis of V oice Quality in Speech Processing,

    E. Keller, “The Analysis of V oice Quality in Speech Processing,” International School on Neural Networks , pp. 54–73, 2004

Show all 34 references
  1. [9]

    GTR- V oice: Articulatory Phonetics Informed Controllable Expressive Speech Synthesis,

    Z. K. Li, M. M. Chen, Y . Zhong, P. Liu, and Z. Duan, “GTR- V oice: Articulatory Phonetics Informed Controllable Expressive Speech Synthesis,” in Interspeech 2024, 2024, pp. 1775–1779

  2. [10]

    PerMod: Perceptually Grounded V oice Modification with Latent Diffusion Mod- els,

    R. Netzorg, A. Jalal, L. McNulty, and G. K. Anumanchipalli, “PerMod: Perceptually Grounded V oice Modification with Latent Diffusion Mod- els,” in 2023 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), 2023, pp. 1–8

  3. [11]

    Consensus Auditory-Perceptual Evaluation of V oice: Development of a Standardized Clinical Protocol,

    G. B. Kempster, B. R. Gerratt, K. V . Abbott, J. Barkmeier-Kraemer, and R. E. Hillman, “Consensus Auditory-Perceptual Evaluation of V oice: Development of a Standardized Clinical Protocol,” 2009

  4. [12]

    ANGUS: Real-Time Manipulation of V ocal Roughness for Emotional Speech Transformations,

    M. Liuni, L. Ardaillon, L. Bonal, L. Seropian, and J.-J. Aucouturier, “ANGUS: Real-Time Manipulation of V ocal Roughness for Emotional Speech Transformations,” arXiv preprint arXiv:2008.11241 , 2020

  5. [13]

    Stochastic Models of Pitch Jitter and Amplitude Shimmer for V oice Modification,

    D. Ruinskiy and Y . Lavner, “Stochastic Models of Pitch Jitter and Amplitude Shimmer for V oice Modification,” in 2008 IEEE 25th Convention of Electrical and Electronics Engineers in Israel , 2008, pp. 489–493

  6. [14]

    Introducing Roughness in Individuality Transformation through Jitter Modeling and Modification,

    A. Verma and A. Kumar, “Introducing Roughness in Individuality Transformation through Jitter Modeling and Modification,” in ICASSP 2005-2005 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) , 2005, pp. 1–5

  7. [15]

    Towards an Interpretable Representation of Speaker Identity via Perceptual V oice Qualities,

    R. Netzorg, B. Yu, A. Guzman, P. Wu, L. McNulty, and G. K. Anumanchipalli, “Towards an Interpretable Representation of Speaker Identity via Perceptual V oice Qualities,” in ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024...

  8. [16]

    Variational Inference with Normalizing Flows,

    D. Rezende and S. Mohamed, “Variational Inference with Normalizing Flows,” in International Conference on Machine Learning. PMLR, 2015, pp. 1530–1538

  9. [17]

    Normalizing Flows: An Introduction and Review of Current Methods,

    I. Kobyzev, S. J. Prince, and M. A. Brubaker, “Normalizing Flows: An Introduction and Review of Current Methods,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 43, no. 11, pp. 3964– 3979, 2020

  10. [18]

    V oiceShop: A Unified Speech-to-Speech Framework for Identity-Preserving Zero-Shot V oice Editing,

    P. Anastassiou, Z. Tang, K. Peng, D. Jia, J. Li, M. Tu, Y . Wang, Y . Wang, and M. Ma, “V oiceShop: A Unified Speech-to-Speech Framework for Identity-Preserving Zero-Shot V oice Editing,” arXiv preprint arXiv:2404.06674, 2024

  11. [19]

    YourTTS: Towards Zero-Shot Multi-Speaker TTS and Zero- Shot V oice Conversion for Everyone,

    E. Casanova, J. Weber, C. D. Shulby, A. C. Junior, E. G ¨olge, and M. A. Ponti, “YourTTS: Towards Zero-Shot Multi-Speaker TTS and Zero- Shot V oice Conversion for Everyone,” in International Conference on Machine Learning. PMLR, 2022, pp. 2709–2720

  12. [20]

    StyleFlow: Attribute- conditioned Exploration of StyleGAN-Generated Images using Condi- tional Continuous Normalizing Flows,

    R. Abdal, P. Zhu, N. J. Mitra, and P. Wonka, “StyleFlow: Attribute- conditioned Exploration of StyleGAN-Generated Images using Condi- tional Continuous Normalizing Flows,” ACM Transactions on Graphics (ToG), vol. 40, no. 3, pp. 1–21, 2021

  13. [21]

    Neural Ordinary Differential Equations,

    R. T. Chen, Y . Rubanova, J. Bettencourt, and D. K. Duvenaud, “Neural Ordinary Differential Equations,” Advances in neural information processing systems, vol. 31, 2018

  14. [22]

    FFJORD: Free-Form Continuous Dynamics for Scalable Reversible Generative Models,

    W. Grathwohl, R. T. Chen, J. Bettencourt, I. Sutskever, and D. Duvenaud, “FFJORD: Free-Form Continuous Dynamics for Scalable Reversible Generative Models,” in International Conference on Learning Repre- sentations, 2019

  15. [23]

    Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech,

    J. Kim, J. Kong, and J. Son, “Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech,” in International Conference on Machine Learning . PMLR, 2021, pp. 5530–5540

  16. [24]

    Frame-Wise and Overlap-Robust Speaker Embeddings for Meeting Diarization,

    T. Cord-Landwehr, C. Boeddeker, C. Zoril ˘a, R. Doddipatla, and R. Haeb- Umbach, “Frame-Wise and Overlap-Robust Speaker Embeddings for Meeting Diarization,” in ICASSP 2023-2023 IEEE International Con- ference on Acoustics, Speech and Signal Processing (ICASSP) , 2023, pp. 1–5

  17. [25]

    LibriTTS-R: A Restored Multi-Speaker Text-to-Speech Corpus,

    Y . Koizumi, H. Zen, S. Karita, Y . Ding, K. Yatabe, N. Morioka, M. Bacchiani, Y . Zhang, W. Han, and A. Bapna, “LibriTTS-R: A Restored Multi-Speaker Text-to-Speech Corpus,” in INTERSPEECH 2023, 2023, pp. 5496–5500

  18. [26]

    LibriTTS: A Corpus Derived from LibriSpeech for Text-to- Speech,

    H. Zen, V . Dang, R. Clark, Y . Zhang, R. J. Weiss, Y . Jia, Z. Chen, and Y . Wu, “LibriTTS: A Corpus Derived from LibriSpeech for Text-to- Speech,” in Interspeech 2019, 2019, pp. 1526–1530

  19. [27]

    Perceptual V oice Qualities Database (PVQD): Database Characteristics,

    P. R. Walden, “Perceptual V oice Qualities Database (PVQD): Database Characteristics,” Journal of Voice, vol. 36, no. 6, pp. 875–e15, 2022

  20. [28]

    HuBERT: Self-Supervised Speech Represen- tation Learning by Masked Prediction of Hidden Units,

    W.-N. Hsu, B. Bolte, Y .-H. H. Tsai, K. Lakhotia, R. Salakhutdinov, and A. rahman Mohamed, “HuBERT: Self-Supervised Speech Represen- tation Learning by Masked Prediction of Hidden Units,” IEEE/ACM Transactions on Audio, Speech, and Language Processing , vol. 29, pp. 3451–3460, 2021

  21. [29]

    WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing,

    S. Chen, C. Wang, Z. Chen, Y . Wu, S. Liu, Z. Chen, J. Li, N. Kanda, T. Yoshioka, X. Xiao, J. Wu, L. Zhou, S. Ren, Y . Qian, Y . Qian, M. Zeng, and F. Wei, “WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing,” IEEE Journal of Selected Topics in Si...

  22. [30]

    Montreal Forced Aligner: Trainable Text-Speech Alignment Using Kaldi,

    M. McAuliffe, M. Socolof, S. Mihuc, M. Wagner, and M. Sonderegger, “Montreal Forced Aligner: Trainable Text-Speech Alignment Using Kaldi,” in Proc. Interspeech 2017 , 2017, pp. 498–502

  23. [31]

    ITU-T Rec. P.808, Subjective evaluation of speech quality with a crowdsourcing approach,

    “ITU-T Rec. P.808, Subjective evaluation of speech quality with a crowdsourcing approach,” 2018

  24. [32]

    V oice Conversion Challenge 2020 — Intra- lingual semi-parallel and cross-lingual voice conversion —,

    Z. Yi, W.-C. Huang, X. Tian, J. Yamagishi, R. K. Das, T. Kinnunen, Z.-H. Ling, and T. Toda, “V oice Conversion Challenge 2020 — Intra- lingual semi-parallel and cross-lingual voice conversion —,” in Joint Workshop for the Blizzard Challenge and Voice Conversion Challenge 2020,...

  25. [33]

    Constructing Confidence Sets Using Rank Statistics,

    D. F. Bauer, “Constructing Confidence Sets Using Rank Statistics,” Journal of the American Statistical Association , vol. 67, no. 339, pp. 687–690, 1972

  26. [34]

    Interactions Between Breathy and Rough V oice Qualities and Their Contributions to Overall Dysphonia Severity,

    Y . Park, S. Anand, L. M. Kopf, R. Shrivastav, and D. A. Eddins, “Interactions Between Breathy and Rough V oice Qualities and Their Contributions to Overall Dysphonia Severity,” Journal of Speech, Language, and Hearing Research, vol. 65, no. 11, pp. 4071–4084, 2022

Pith tools

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