Pith. sign in

REVIEW 5 major objections 5 minor 34 references

I2TTS: Image-indicated Immersive Text-to-speech Synthesis with Spatial Perception

T0 review · 5 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read I2TTS is an end-to-end multimodal TTS framework that synthesizes speech whose reverberation matches a visual scene prompt, while preserving naturalness and supporting zero-shot voice adaptation.

desk verdict Plausible incremental architecture, but the central spatial-matching claim is unverified because SRE is undefined and may be computed with the same frozen classifier used as a training loss. read the letter →

arxiv 2411.13314 v4 pith:HL2NG5V4 submitted 2024-11-20 cs.SD eess.AS

Jiawei Zhang , Tian-Hao Zhang , Jun Wang , Jiaran Gao , Xinyuan Qian , Xu-Cheng Yin This is my paper · ORCID
classification cs.SDeess.AS
keywords text-to-speechspatialaudioreverberationvisualscenepromptmultimodalTTSzero-shotspeakeradaptationCLIPVITS
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 claims that text-to-speech can be made spatially aware by generating reverberant speech directly from a text input plus a photo of the acoustic scene, rather than applying room acoustics as a post-processing step. The proposed system, I2TTS, feeds CLIP visual features through an adapter into a VITS backbone, uses a frozen speech-reverberation classifier as a training-time refinement loss, and adds a speaker encoder for zero-shot voice adaptation. The authors report that this end-to-end design beats the earlier image-conditioned systems ViT-TTS and MS2KU-VTTS on word error rate, spectral distance, speaker similarity, scene recognition error, and listener ratings, while keeping naturalness intact. A sympathetic reader would take the central claim to be that scene-aware reverberation can be learned jointly with content and speaker identity without degrading intelligibility or voice quality.

What carries the argument

The load-bearing machinery is a VITS text-to-speech backbone that receives two extra conditionings: a scene embedding obtained by passing the scene image through a CLIP image encoder and an MLP adapter, and a speaker embedding obtained from a short reference clip with a StyleEncoder-style module. A Speech Reverberation Classifier (SRC), made of four 2-D convolution blocks plus a transformer block and a classification head, is pretrained on the same convolved training speech, frozen, and used to compute a cross-entropy loss on the generated mel-spectrogram so the decoder learns to produce reverberation consistent with the image. The adapter is also guided with a cross-entropy loss against the reverb category, tying the visual embedding to the acoustic label space.

What would settle it

Compute the paper's Space Recognition Error using a room classifier trained on held-out room impulse responses that were never mixed into the training set, and compare the same generated samples against ViT-TTS; if the 27.2%-versus-64.2% gap shrinks or disappears, the spatial-matching claim is not established. A complementary check is a forced-choice listening test where raters pick which of several room photos matches each generated audio clip.

Watch

Extended reading notes

Core claim

The central claim is that I2TTS synthesizes high-quality immersive speech from text and visual scene prompts in one end-to-end pipeline, producing speech whose reverberation matches the image's room acoustics. Built on a VITS backbone, the model injects a CLIP-derived scene embedding and a speaker embedding into the generative flow, and a frozen Speech Reverberation Classifier refines the generated mel-spectrogram by predicting the reverb class and feeding a cross-entropy loss back during training. In the reported evaluations the full model reaches a 27.2% space recognition error against 64.2% for ViT-TTS and 57.1% for MS2KU-VTTS, with the best WER (7.6%), MCD (4.22), SECS (0.62), and IMOS (3.96), which the authors take as evidence that scene and spatial matching can be achieved without compromising naturalness.

Load-bearing premise

The main evidence that the synthesized speech matches the scene rests on a Space Recognition Error metric that the paper never defines, while the classifier used to train the model is pretrained on the same convolved speech, so the reported drop in SRE could partly reflect the model learning to fool that classifier.

Editorial extensions

If this is right

  • Because the decoder outputs the waveform directly, the method avoids the external vocoder that previous image-conditioned TTS systems required, removing a source of synthesis artifacts.
  • A scene photo becomes a practical control signal: the same text can be rendered with the acoustics of a cathedral, a hallway, or a recording booth from the corresponding image.
  • The speaker encoder allows an unseen voice to be placed into a specified acoustic scene from a few seconds of reference speech, enabling personalized VR and gaming audio.
  • The SRC refinement appears to help in both wide and narrow environments, with reported SRE of 25.3% and 28.6% respectively, suggesting the scene conditioning does not overfit to one acoustic scale.

Reading between the lines

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

  • Extension: the same scene-embedding adapter could be paired with different acoustic-transformation targets, such as head-related transfer functions for binaural direction, making spatial TTS extend from room reverb to full three-dimensional audio.
  • Extension: because the frozen classifier used to train the model was pretrained on the same convolved audio that forms the training data, the reported SRE gains should be re-checked with a classifier trained on held-out room impulse responses or with human scene-matching tests.
  • Extension: feeding the SRC with the ground-truth reference speech as a contrastive negative could turn the refinement stage into a true alignment mechanism rather than a soft label predictor.
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

5 major / 5 minor

Summary. The paper presents I2TTS, an end-to-end multispeaker text-to-speech system that takes a text input, a visual scene prompt, and an optional speaker reference to synthesize reverberant speech. The architecture combines a CLIP-based image encoder with an adapter, a speaker encoder for zero-shot voice cloning, and a frozen Speech Reverberation Classifier (SRC) used during training to refine acoustic-scene alignment, all built on a VITS backbone. Evaluations on LJSpeech, VCTK, and Image2Reverb report improved word error rate, mel cepstral distortion, speaker similarity, and especially a large reduction in Space Recognition Error (SRE) compared with ViT-TTS and MS2KU-VTTS, alongside subjective NMOS/SMOS/IMOS scores.

Significance. If the reported results hold, the paper makes a useful contribution to multi-modal TTS by integrating visual scene conditioning directly into an end-to-end VITS-based pipeline and by demonstrating zero-shot speaker adaptation. The architectural direction is reasonable and the ablation variants provide some evidence for the usefulness of the CLIP encoder and the SRC module. However, the central claim of spatially accurate scene-reverb matching rests on an undefined metric and a potentially circular evaluation protocol. The paper also provides no statistical significance measures for its objective metrics and only a very small subjective evaluation. These issues currently prevent the claimed superiority from being accepted, although they are addressable in a revision.

major comments (5)
  1. [§IV-B and Table II] SRE is never formally defined. The text only states that 'SRE measures the correctness of the room acoustics of the generated speech,' and Table II inconsistently calls it 'Scene Reverb Error.' The paper does not specify the classifier used for evaluation, the set of reverb classes, how ground-truth labels are assigned, or how errors are aggregated. Since the reported drop from 64.2% (ViT-TTS) to 27.2% (Proposed) is the main objective evidence for the spatial-matching claim, the metric must be defined precisely for the results to be verifiable.
  2. [§IV-A and §III-C] The evaluation of SRE risks being circular. The SRC is pretrained on the same convolved speech data used for training the TTS system, is frozen during TTS training, and is used to compute a classification loss on generated speech. If SRE is computed with this same SRC or a substantially similar classifier, then the model is directly optimizing the evaluation metric, and the improvement over baselines would reflect overfitting to the SRC's decision boundary rather than generalizable acoustic-scene matching. The paper must either use an independent held-out classifier for SRE evaluation or explicitly report the relationship between the training-time classifier and the evaluation classifier.
  3. [§IV-C, Table I] Objective metrics are reported without error bars, confidence intervals, or statistical significance tests. For example, WER differences between the proposed method (7.6) and the w/o CLIP ablation (7.8) are small, and SECS differences (0.62 vs. 0.57) may be within normal variation. Without repeated runs or significance testing, the claim of 'consistently superior performance' across objective metrics is not supported.
  4. [§IV-B] The subjective IMOS evaluation is described as engaging 20 listeners each evaluating 6 test samples, but the paper does not state whether each listener rated all systems on the same samples, how many judgments per condition were collected, or whether the reported standard deviations (<0.25) are across listeners or across samples. This is too small a basis to support the strong IMOS claims in Table I and Table II, especially the environment-wise IMOS differences of 0.15-0.3.
  5. [§III-C] The role of the SRC is ambiguous. The method section says the SRC 'refines' the mel-spectrogram and 'adjusts' reverberation, but the only described use is computing a cross-entropy loss on the predictions of the generated speech. No forward-pass post-processing or refinement equation is given. The contribution summary in §I also alternately calls it a 'Scene Relevance Classifier' and a 'Speech Reverberation Classifier,' which makes the intended mechanism unclear. Please clarify whether the SRC acts only as a training-time regularizer or performs any inference-time correction.
minor comments (5)
  1. [§III-C] There is a typo in 'reveives the mel-spectrogram' that should be 'receives.'
  2. [§IV-B] The metric is named 'Space Recognition Error' in §IV-B but 'Scene Reverb Error' in Table II; please use one consistent name.
  3. [§IV-A] The description of the baseline is ambiguous: 'the VITS backbone trained on the clean speech followed by convolution with RIR' could mean the generated clean speech is convolved at inference time, but it is unclear whether the baseline sees any scene information. Please specify.
  4. [§IV-E] The Grad-CAM visualization in Figure 2 is qualitative and does not demonstrate that the highlighted image regions are acoustically relevant. Consider quantifying the correlation between attention maps and acoustic parameters such as RT60.
  5. [General] No code or demo audio samples are provided in the manuscript; releasing at least the evaluation code for SRE computation would greatly help reproducibility.

Circularity Check

1 steps flagged · score 6.0 of 10

Headline SRE metric is undefined and may be computed by the same frozen SRC classifier used as a training loss, so the central spatial-matching improvement is not independently established.

  1. fitted input called prediction [Section IV-A Implementation Details and Section IV-B Evaluation Metrics]
    "Additionally, we pretrained the SRC model on the convolutional speech samples over 300 epochs. Once integrated into the TTS framework, we froze its parameters and only used the model to calculate the classification loss of the generated speech. ... SRE measures the correctness of the room acoustics of the generated speech."

    The paper's only reverb classifier, SRC, is pretrained on the same convolved-speech samples, frozen, and used as the classification loss for generated speech. SRE is never defined; the sole description is that it 'measures the correctness of the room acoustics of the generated speech,' the same task the frozen SRC performs. If SRE uses the SRC, the reported drop from 64.2% to 27.2% is the model being scored by the very classifier it was trained to satisfy, while ViT-TTS was not trained with that classifier. No independent SRE pipeline, evaluation classifier, or code is provided, so this reduction of the headline metric to the training objective cannot be ruled out.

full rationale

The central spatial-matching claim rests on the objective SRE improvement. The only described reverb classifier, SRC, is pretrained on the same convolved speech, frozen, and used as a training loss for generated speech; SRE is described only as measuring room-acoustic correctness, with no evaluation protocol specified. If the evaluation uses that classifier (or one derived from it), the improvement reduces to matching the training classifier. WER, MCD, SECS, and the subjective NMOS/SMOS/IMOS are independent and not circular; no load-bearing self-citations exist. But because the paper's primary spatial-matching evidence is not verifiably independent of its training objective, partial circularity score 6 is warranted.

Assumptions & free parameters 1 free parameters · 3 assumptions · 1 invented entities

The central claim depends on the transferability of CLIP features to acoustics, the validity of Image2Reverb as ground truth, the realism of convolutional reverberation, and the SRC loss. The SRC itself carries a circularity risk because the evaluation metric SRE is undefined and may coincide with the SRC.

free parameters (1)
  • SRC classification loss weight = not reported
    The loss combining the SRC classification error with VITS losses is not specified, which affects training behavior and reproducibility.
assumptions (3)
  • domain assumption CLIP features pretrained on image-text data transfer to acoustic-scene properties
    The image encoder is a frozen CLIP model; the paper provides no independent validation that its representations correlate with reverberation beyond the task-specific adapters.
  • domain assumption Image2Reverb dataset provides reliable image-to-RIR pairs
    The supervised pairs of scene images and room impulse responses are taken as ground truth for training the reverb classifier and TTS.
  • domain assumption Convolving clean speech with measured RIRs produces realistic reverberant speech suitable as ground truth
    The training targets are generated by random convolution; if these are not natural-sounding or are mispaired, the model learns the wrong mapping.
invented entities (1)
  • Speech Reverberation Classifier (SRC)
    purpose: Classifies the reverberation category of synthesized mel-spectrograms and provides a training loss to align generated audio with the input scene.
    The SRC is pretrained on the same convolved data and frozen during TTS training; its role in the headline SRE metric is not disclosed, so its independence as an evaluation signal is questionable.

how reviews work

0 comments
Cite this review

Pith. "Pith review of I2TTS: Image-indicated Immersive Text-to-speech Synthesis with Spatial Perception." pith.science (2026). https://pith.science/paper/HL2NG5V4

@misc{pith2026241113314,
  author       = {Pith},
  title        = {Pith review of: I2TTS: Image-indicated Immersive Text-to-speech Synthesis with Spatial Perception},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HL2NG5V4}},
  note         = {Machine review of arXiv:2411.13314}
}
read the original abstract

Controlling the style and characteristics of speech synthesis is crucial for adapting the output to specific contexts and user requirements. Previous Text-to-speech (TTS) works have focused primarily on the technical aspects of producing natural-sounding speech, such as intonation, rhythm, and clarity. However, they overlook the fact that there is a growing emphasis on spatial perception of synthesized speech, which may provide immersive experience in gaming and virtual reality. To solve this issue, in this paper, we present a novel multi-modal TTS approach, namely Image-indicated Immersive Text-to-speech Synthesis (I2TTS). Specifically, we introduce a scene prompt encoder that integrates visual scene prompts directly into the synthesis pipeline to control the speech generation process. Additionally, we propose a reverberation classification and refinement technique that adjusts the synthesized mel-spectrogram to enhance the immersive experience, ensuring that the involved reverberation condition matches the scene accurately. Experimental results demonstrate that our model achieves high-quality scene and spatial matching without compromising speech naturalness, marking a significant advancement in the field of context-aware speech synthesis. Project demo page: https://spatialTTS.github.io/ Index Terms-Speech synthesis, scene prompt, spatial perception

Figures

Figures reproduced from arXiv: 2411.13314 by the authors.

Figure 1
Figure 1. Block diagram of the I2TTS model (⊕ indicates element-wise addition). of the scene into the speech synthesis pipeline. This module is simply comprised of a MLP network. In practice, we first input the scene image Ii into the CLIP￾based image encoder to extract the scene prompt features. After being fed to the adapter, we transform the scene prompt features into an acoustic space embedding Ie. To make this embedding … view at source ↗
Figure 2
Figure 2. Gradient-weighted Class Activation Mapping (Grad-CAM)s for [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 23 canonical work pages

  1. [1]

    Tacotron: Towards end-to- end speech synthesis,

    Y . Wang, R. Skerry-Ryan, D. Stanton, Y . Wu, R. J. Weiss, N. Jaitly, Z. Yang, Y . Xiao, Z. Chen, S. Bengio et al., “Tacotron: Towards end-to- end speech synthesis,” arXiv preprint arXiv:1703.10135 , 2017

    arXiv 2017

  2. [2]

    Fastspeech: Fast, robust and controllable text to speech,

    Y . Ren, Y . Ruan, X. Tan, T. Qin, S. Zhao, Z. Zhao, and T.-Y . Liu, “Fastspeech: Fast, robust and controllable text to speech,” Advances in Neural Information Processing Systems , vol. 32, 2019

    work page 2019

  3. [3]

    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 Proc. of Int. Conf. on Machine Learning . PMLR, 2021, pp. 5530–5540

    work page 2021

  4. [4]

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

    C. Wang, S. Chen, Y . Wu, Z. Zhang, L. Zhou, S. Liu, Z. Chen, Y . Liu, H. Wang, J. Li et al., “Neural codec language models are zero-shot text to speech synthesizers,” arXiv preprint arXiv:2301.02111 , 2023

    arXiv 2023

  5. [5]

    Yourtts: Towards zero-shot multi-speaker tts and zero-shot voice 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 voice conversion for everyone,” in Proc. of Int. Conf. on Machine Learning . PMLR, 2022, pp. 2709–2720

    work page 2022

  6. [6]

    Prompttts: Controllable text-to-speech with text descriptions,

    Z. Guo, Y . Leng, Y . Wu, S. Zhao, and X. Tan, “Prompttts: Controllable text-to-speech with text descriptions,” in Proc. of IEEE Int. Conf. on Audio, Speech and Signal Processing . IEEE, 2023, pp. 1–5

    work page 2023

  7. [7]

    Instructtts: Mod- elling expressive tts in discrete latent space with natural language style prompt,

    D. Yang, S. Liu, R. Huang, C. Weng, and H. Meng, “Instructtts: Mod- elling expressive tts in discrete latent space with natural language style prompt,” IEEE/ACM Trans. on Audio, Speech and Language Processing, 2024. 5

    work page 2024

  8. [8]

    Mm-tts: Multi-modal prompt based style transfer for expres- sive text-to-speech synthesis,

    W. Guan, Y . Li, T. Li, H. Huang, F. Wang, J. Lin, L. Huang, L. Li, and Q. Hong, “Mm-tts: Multi-modal prompt based style transfer for expres- sive text-to-speech synthesis,” in AAAI Conf. on Artificial Intelligence , vol. 38, no. 16, 2024, pp. 18 117–18 125

    work page 2024

Show all 34 references
  1. [9]

    Comparison of different impulse response measurement techniques,

    G.-B. Stan, J.-J. Embrechts, and D. Archambeau, “Comparison of different impulse response measurement techniques,” Journal of the Audio Engineering Society , vol. 50, no. 4, pp. 249–262, 2002

    2002

  2. [10]

    Environment aware text-to-speech synthesis,

    D. Tan, G. Zhang, and T. Lee, “Environment aware text-to-speech synthesis,” arXiv preprint arXiv:2110.03887 , 2021

    2021 arXiv

  3. [11]

    V oiceldm: Text-to-speech with environmental context,

    Y . Lee, I. Yeon, J. Nam, and J. S. Chung, “V oiceldm: Text-to-speech with environmental context,” in ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE, 2024, pp. 12 566–12 571

    2024

  4. [12]

    Vit-tts: visual text-to-speech with scalable diffusion transformer,

    H. Liu, R. Huang, X. Lin, W. Xu, M. Zheng, H. Chen, J. He, and Z. Zhao, “Vit-tts: visual text-to-speech with scalable diffusion transformer,” arXiv preprint arXiv:2305.12708, 2023

    2023 arXiv

  5. [13]

    Multi-source spatial knowledge understanding for immersive visual text-to-speech,

    S. He and R. Liu, “Multi-source spatial knowledge understanding for immersive visual text-to-speech,” in ICASSP 2025-2025 IEEE Interna- tional Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2025, pp. 1–5

    2025

  6. [14]

    Learning transferable visual models from natural language supervision,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark et al., “Learning transferable visual models from natural language supervision,” in Proc. of Int. Conf. on Machine Learning . PMLR, 2021, pp. 8748–8763

    2021

  7. [15]

    Allen, M

    J. Allen, M. S. Hunnicutt, D. H. Klatt, R. C. Armstrong, and D. B. Pisoni, From text to speech: The MITalk system . Cambridge University Press, 1987

    1987

  8. [16]

    The festival speech synthesis system,

    A. Black, P. Taylor, R. Caley, and R. Clark, “The festival speech synthesis system,” 1998

    1998

  9. [17]

    An hmm-based speech synthesis system applied to english,

    K. Tokuda, H. Zen, and A. W. Black, “An hmm-based speech synthesis system applied to english,” inIEEE speech synthesis workshop. Citeseer, 2002, pp. 227–230

    2002

  10. [18]

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

    Y . Ren, C. Hu, X. Tan, T. Qin, S. Zhao, Z. Zhao, and T.-Y . Liu, “Fastspeech 2: Fast and high-quality end-to-end text to speech,” arXiv preprint arXiv:2006.04558, 2020

    2006 arXiv

  11. [19]

    Diff- tts: A denoising diffusion model for text-to-speech,

    M. Jeong, H. Kim, S. J. Cheon, B. J. Choi, and N. S. Kim, “Diff- tts: A denoising diffusion model for text-to-speech,” arXiv preprint arXiv:2104.01409, 2021

    2021 arXiv

  12. [20]

    Prodiff: Progressive fast diffusion model for high-quality text-to-speech,

    R. Huang, Z. Zhao, H. Liu, J. Liu, C. Cui, and Y . Ren, “Prodiff: Progressive fast diffusion model for high-quality text-to-speech,” inProc. of ACM Int. Conf. on Multimedia , 2022, pp. 2595–2605

    2022

  13. [21]

    Clam-tts: Improving neural codec language model for zero-shot text-to-speech,

    J. Kim, K. Lee, S. Chung, and J. Cho, “Clam-tts: Improving neural codec language model for zero-shot text-to-speech,” in Proc. ofInt. Conf. on Learning Representations, 2023

    2023

  14. [22]

    Audiolm: a language modeling approach to audio generation,

    Z. Borsos, R. Marinier, D. Vincent, E. Kharitonov, O. Pietquin, M. Shar- ifi, D. Roblek, O. Teboul, D. Grangier, M. Tagliasacchi et al., “Audiolm: a language modeling approach to audio generation,” IEEE/ACM Trans. on Audio, Speech and Language Processing , 2023

    2023

  15. [23]

    Naturalspeech 3: Zero-shot speech syn- thesis with factorized codec and diffusion models,

    Z. Ju, Y . Wang, K. Shen, X. Tan, D. Xin, D. Yang, Y . Liu, Y . Leng, K. Song, S. Tang et al. , “Naturalspeech 3: Zero-shot speech syn- thesis with factorized codec and diffusion models,” arXiv preprint arXiv:2403.03100, 2024

    2024 arXiv

  16. [24]

    Image2reverb: Cross-modal reverb impulse response synthesis,

    N. Singh, J. Mentch, J. Ng, M. Beveridge, and I. Drori, “Image2reverb: Cross-modal reverb impulse response synthesis,” in Proc. of Int. Conf. on Computer Vision , 2021, pp. 286–295

    2021

  17. [25]

    Visual acoustic match- ing,

    C. Chen, R. Gao, P. Calamia, and K. Grauman, “Visual acoustic match- ing,” in Proc. of Int. Conf. on Computer Vision and Pattern Recognition , 2022, pp. 18 858–18 868

    2022

  18. [26]

    Self-supervised visual acoustic matching,

    A. Somayazulu, C. Chen, and K. Grauman, “Self-supervised visual acoustic matching,” Advances in Neural Information Processing Systems, vol. 36, 2024

    2024

  19. [27]

    Meta-stylespeech: Multi- speaker adaptive text-to-speech generation,

    D. Min, D. B. Lee, E. Yang, and S. J. Hwang, “Meta-stylespeech: Multi- speaker adaptive text-to-speech generation,” in International Conference on Machine Learning . PMLR, 2021, pp. 7748–7759

    2021

  20. [28]

    Attention is all you need,

    A. Vaswani, “Attention is all you need,” Advances in Neural Information Processing Systems, 2017

    2017

  21. [29]

    The lj speech dataset,

    K. Ito and L. Johnson, “The lj speech dataset,” 2017

    2017

  22. [30]

    Cstr vctk corpus: English multi-speaker corpus for cstr voice cloning toolkit,

    C. Veaux, J. Yamagishi, K. MacDonald et al. , “Cstr vctk corpus: English multi-speaker corpus for cstr voice cloning toolkit,”University of Edinburgh. The Centre for Speech Technology Research (CSTR) , vol. 6, p. 15, 2017

    2017

  23. [31]

    Mel-cepstral distance measure for objective speech quality assessment,

    R. Kubichek, “Mel-cepstral distance measure for objective speech quality assessment,” in Proceedings of IEEE pacific rim conference on commu- nications computers and signal processing , vol. 1. IEEE, 1993, pp. 125–128

    1993

  24. [32]

    Sc-glowtts: An efficient zero-shot multi-speaker text-to-speech model,

    E. Casanova, C. Shulby, E. G ¨olge, N. M. M ¨uller, F. S. De Oliveira, A. C. Junior, A. d. S. Soares, S. M. Aluisio, and M. A. Ponti, “Sc-glowtts: An efficient zero-shot multi-speaker text-to-speech model,” arXiv preprint arXiv:2104.05557, 2021

    2021 arXiv

  25. [33]

    An overview of voice con- version and its challenges: From statistical modeling to deep learning,

    B. Sisman, J. Yamagishi, S. King, and H. Li, “An overview of voice con- version and its challenges: From statistical modeling to deep learning,” IEEE/ACM Trans. on Audio, Speech and Language Processing , vol. 29, pp. 132–157, 2020

    2020

  26. [34]

    Grad-cam: Visual explanations from deep networks via gradient-based localization,

    R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-cam: Visual explanations from deep networks via gradient-based localization,” in Proc. of Int. Conf. on Computer Vision , 2017, pp. 618–626. 6

    2017

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.