CycleGAN-VC: Non-parallel Voice Conversion Using Cycle-Consistent Adversarial Networks
The DOI matches its Crossref record; 60 of the 61 references it cites were found in a registry or an index.
Abstract
We propose a non-parallel voice-conversion (VC) method that can learn a mapping from source to target speech without relying on parallel data. The proposed method is particularly noteworthy in that it is general purpose and high quality and works without any extra data, modules, or alignment procedure. Our method, called CycleGAN-VC, uses a cycle-consistent adversarial network (CycleGAN) with gated convolutional neural networks (CNNs) and an identity-mapping loss. A CycleGAN learns forward and inverse mappings simultaneously using adversarial and cycle-consistency losses. This makes it possible to find an optimal pseudo pair from non-parallel data. Furthermore, the adversarial loss can bring the converted speech close to the target one on the basis of indistinguishability without explicit density estimation. This allows to avoid over-smoothing caused by statistical averaging, which occurs in many conventional statistical model-based VC methods that represent data distribution explicitly. We configure a CycleGAN with gated CNNs and train it with an identity-mapping loss. This allows the mapping function to capture sequential and hierarchical structures while preserving linguistic information. We evaluated our method on a non-parallel VC task. An objective evaluation showed that the converted feature sequence was near natural in terms of global variance and modulation spectra, which are structural indicators highly correlated with subjective evaluation. A subjective evaluation showed that the quality of the converted speech was comparable to that obtained with a Gaussian mixture model-based parallel VC method even though CycleGAN-VC is trained under disadvantageous conditions (non-parallel and half the amount of data).
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Recent research [59,60,61,62,63] in VC technologies has addressed long-standing challenges such as data scarcity, cross-lingual conversion, and real-time processing.”
From Audio Deepfake Detection: What Has Been Achieved and What Lies Ahead (2025) · cites this paper as [59]
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 1×
“Inspired by a previous study [49] for neural style transfer and super-resolution, we used the network that included downsampling, residual [50], and upsampling layers, as well as incorporating instance normalization [51].”
MethodAdam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · cited 1×
“We trained the network using the Adam optimizer [53] with a batch size of 1.”
MethodFully convolutional networks for semantic segmentation
Long, J., Shelhamer, E., & Darrell, T. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298965 · Cited by 38,834
Unpaired image-to-image translation using cycle-consistent adversarial networks
Zhu, J.-Y., Park, T., Isola, P., & Efros, A. A. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.244 · Cited by 23,602
Labaca-Castro, R. (2023) · Machine Learning under Malware Attack · doi:10.1007/978-3-658-40442-0_9 · Cited by 19,272
GAN(Generative Adversarial Nets)
(2017) · Journal of Japan Society for Fuzzy Theory and Intelligent Informatics · doi:10.3156/jsoft.29.5_177_2 · Cited by 19,270
(2010) · International Conference on Machine Learning · Cited by 13,995
Perceptual Losses for Real-Time Style Transfer and Super-Resolution
Johnson, J., Alahi, A., & Fei-Fei, L. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46475-6_43 · Cited by 11,283 · cited 2×
“Inspired by a previous study [49] for neural style transfer and super-resolution, we used the network that included downsampling, residual [50], and upsampling layers, as well as incorporating instance normalization [51].”
MethodShi, W., Caballero, J., Huszar, F., Totz, J., Aitken, A. P., Bishop, R., Rueckert, D., & Wang, Z. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.207 · Cited by 7,570
“Least squares generative adversarial networks,”
Mao, X., Li, Q., Xie, H., Lau, R. Y. K., Wang, Z., & Smolley, S. P. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.304 · Cited by 5,437 · cited 1×
“To stabilize training, we used a least squares GAN [52], which replaces the negative log likelihood objective in Ladv by a least squares loss.”
Method
Research works printed but not linked: 1
- maas, 2013, Rectifier nonlinearities improve neural network acoustic models, Proc ICML Workshop
Cited by
A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
Gui, J., Sun, Z., Wen, Y., Tao, D., & Ye, J. (2023) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2021.3130191 · Cited by 1,231
Masood, M., Nawaz, M., Malik, K. M., Javed, A., Irtaza, A., & Malik, H. (2023) · Applied Intelligence · doi:10.1007/s10489-022-03766-z · Cited by 543
“Recent GAN-based approaches, such as CycleGAN [253–256], VAW-GAN [257], and StarGAN [258], attempt to achieve high-quality transformed speech using non-parallel training data.”
MethodAn Overview of Voice Conversion and Its Challenges: From Statistical Modeling to Deep Learning
Sisman, B., Yamagishi, J., King, S., & Li, H. (2021) · IEEE/ACM Transactions on Audio, Speech, and Language Processing · doi:10.1109/taslp.2020.3038524 · Cited by 357
“Combining these three loss functions, we can obtain the overall loss function of CycleGAN [48], [210].”
BackgroundMethodCycleGAN-VC2: improved CycleGAN-based non-parallel voice conversion.
Kaneko, T., Kameoka, H., Tanaka, K., & Hojo, N. (2019) · ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) · doi:10.1109/icassp.2019.8682897 · Cited by 272
Self-citationEmotional voice conversion: Theory, databases and ESD
Zhou, K., Sisman, B., Liu, R., & Li, H. (2022) · Speech Communication · doi:10.1016/j.specom.2021.11.006 · Cited by 181
Deepfake generation and detection: Case study and challenges
Patel, Y., Tanwar, S., Gupta, R., Bhattacharya, P., Davidson, I. E., Nyameko, R., Aluvala, S., & Vimal, V. (2023) · IEEE Access · doi:10.1109/access.2023.3342107 · Cited by 186
“The model is often compared to CycleGAN-VC [71], which has a limited one-to-one mapping capability, compared to many-to-many of StarGAN.”
ResultStarGAN-VC2: Rethinking Conditional Methods for StarGAN-Based Voice Conversion
Kaneko, T., Kameoka, H., Tanaka, K., & Hojo, N. (2019) · Interspeech (USB) · doi:10.21437/interspeech.2019-2236 · Cited by 200
Self-citationACVAE-VC: Non-Parallel Voice Conversion With Auxiliary Classifier Variational Autoencoder
Kameoka, H., Kaneko, T., Tanaka, K., & Hojo, N. (2019) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2019.2917232 · Cited by 181
“(CycleGAN) [31], which was originally proposed as a method for translating images using unpaired training examples [32]– [34].”
MethodSelf-citationA Review of Synthetic Image Data and Its Use in Computer Vision
Man, K., & Chahl, J. (2022) · Journal of Imaging · doi:10.3390/jimaging8110310 · Cited by 172
“CycleGAN has seen some derivative models, with most adapting it for use in specific applications, such as CycleGAN-VC [51–53] for voice conversion, res-cycle GAN [54] for generating medical scan pairs, and Cycle-Dehaze [55], a variant for dehazing images.”
MethodVoice Conversion Challenge 2020 –- Intra-lingual semi-parallel and cross-lingual voice conversion –-
Yi, Z., Huang, W.-C., Tian, X., Yamagishi, J., Das, R. K., Kinnunen, T., Ling, Z.-H., & Toda, T. (2020) · Joint Workshop for the Blizzard Challenge and Voice Conversion Challenge 2020 · doi:10.21437/vcc_bc.2020-14 · Cited by 126
Again-VC: A One-Shot Voice Conversion Using Activation Guidance and Adaptive Instance Normalization
Chen, Y.-H., Wu, D.-Y., Wu, T.-H., & Lee, H.-Y. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9414257 · Cited by 147
“, to tackle many-to-many voice conversion [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14].”
Zhang, J.-X., Ling, Z.-H., & Dai, L.-R. (2019) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2019.2960721 · Cited by 127
“nearest neighbor search step and a conversion step alignment (INCA) [23], [24] and CycleGAN-based VC [25]–[27].”
MethodCycleGAN-VC3: Examining and Improving CycleGAN-VCs for Mel-Spectrogram Conversion
Kaneko, T., Kameoka, H., Tanaka, K., & Hojo, N. (2020) · Interspeech (USB) · doi:10.21437/interspeech.2020-2280 · Cited by 124
Self-citationAn Overview of Affective Speech Synthesis and Conversion in the Deep Learning Era
Triantafyllopoulos, A., Schuller, B. W., İymen, G., Sezgin, T. M., He, X., Yang, Z., Tzirakis, P., Liu, S., Mertes, S., André, E., et al. (2023) · Proceedings of the IEEE · doi:10.1109/jproc.2023.3250266 · Cited by 150
“However, despite several attempts, no notable progress was made until recent years, which saw the advent of DNNs. Recently proposed methods exploit the representation power of DNNs by means of VAEs [92], [93], GANs [94], [95], and seq2seq models [96], [97].”
BackgroundMethodLuleci, F., Çatbaş, F. N., & Avcı, O. (2023) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2023.110370 · Cited by 105
MBNET: MOS Prediction for Synthesized Speech with Mean-Bias Network
Leng, Y., Tan, X., Zhao, S., Soong, F. K., Li, X., & Qin, T. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9413877 · Cited by 140
“Speech quality assessment [1] aims to measure the quality of synthesized speech and has long been a challenge in speech synthesis areas such as text to speech [2, 3, 4] and voice conversion [5, 6].”
BackgroundUnderstanding GANs: fundamentals, variants, training challenges, applications, and open problems
Ahmad, Z., Jaffri, Z. U. A., Chen, M., & Bao, S. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-19361-y · Cited by 108
Any-to-Many Voice Conversion With Location-Relative Sequence-to-Sequence Modeling
Liu, S., Cao, Y., Wang, D., Wu, X., Liu, X., & Meng, H. M. L. (2021) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2021.3076867 · Cited by 145
VQVC+: One-Shot Voice Conversion by Vector Quantization and U-Net Architecture
Wu, D.-Y., Chen, Y.-H., & Lee, H.-Y. (2020) · Interspeech 2020 · doi:10.21437/interspeech.2020-1443 · Cited by 88
Generative adversarial networks for speech processing: A review
Wali, A., Alamgir, Z., Karim, S., Fawaz, A., Ali, M. B., Adan, M., & Mujtaba, M. (2021) · Computer Speech & Language · doi:10.1016/j.csl.2021.101308 · Cited by 100
Shaaban, O. A., Yıldırım, R., & Alguttar, A. A. (2023) · IEEE Access · doi:10.1109/access.2023.3333866 · Cited by 107
“Robust value creation approaches, such as neural network-based [108], vocoder [109], [110], GAN [111], [112], and VAE [113], [114] have been developed to aid in the modeling of non-parallel spectral data.”
BackgroundMethodTransforming Spectrum and Prosody for Emotional Voice Conversion with Non-Parallel Training Data
Zhou, K., Şişman, B., & Li, H. (2020) · The Speaker and Language Recognition Workshop (Odyssey) · doi:10.21437/odyssey.2020-33 · Cited by 80
Maskcyclegan-VC: Learning Non-Parallel Voice Conversion with Filling in Frames
Kaneko, T., Kameoka, H., Tanaka, K., & Hojo, N. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9414851 · Cited by 91
Self-citationTransfer Learning From Speech Synthesis to Voice Conversion With Non-Parallel Training Data
Zhang, M., Zhou, Y., Li, Z., & Li, H. (2021) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2021.3066047 · Cited by 90
ConvS2S-VC: Fully Convolutional Sequence-to-Sequence Voice Conversion
Kameoka, H., Tanaka, K., Kwaśny, D., Kaneko, T., & Hojo, N. (2020) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2020.3001456 · Cited by 75
Self-citationAudio Deepfake Detection: What Has Been Achieved and What Lies Ahead
Zhang, B., Cui, H., Nguyen, V., & Whitty, M. (2025) · Sensors · doi:10.3390/s25071989 · Cited by 72
“Recent research [59–63] in VC technologies has addressed long-standing challenges such as data scarcity, cross-lingual conversion, and real-time processing.”
Lin, Y. Y., Chien, C.-M., Lin, J.-H., Lee, H.-Y., & Lee, L.-S. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9413699 · Cited by 96
“No additional loss term is needed as done previously [3, 4, 5].”
BackgroundMethodDomain Generalization via Aggregation and Separation for Audio Deepfake Detection
Xie, Y., Cheng, H., Wang, Y., & Ye, L. (2024) · IEEE Transactions on Information Forensics and Security · doi:10.1109/tifs.2023.3324724 · Cited by 59
Avqvc: One-Shot Voice Conversion By Vector Quantization With Applying Contrastive Learning
Tang, H., Zhang, X., Wang, J., Cheng, N., & Xiao, J. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9746369 · Cited by 74
NVC-Net: End-To-End Adversarial Voice Conversion
Nguyen, B., & Cardinaux, F. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9747020 · Cited by 56
Wang, M., Xie, P., Du, Y., & Hu, X. (2023) · Applied Sciences · doi:10.3390/app13127111 · Cited by 46
“Based on CycleGAN, different models have been proposed for face transfer [25], Chinese handwritten character generation [26], image generation from text [27], image correction [28], and tasks in the audio field [29–31].”
MethodMelGAN-VC: Voice Conversion and Audio Style Transfer on arbitrarily long samples using SpectrogramsPreprint
Pasini, M. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.03713 · Cited by 44
Chan, C. H., Qian, K., Zhang, Y., & Hasegawa‐Johnson, M. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9747763 · Cited by 73
The Impact of Silence on Speech Anti-Spoofing
Zhang, Y., Li, Z., Lu, J., Hua, H., Wang, W., & Zhang, P. (2023) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2023.3306711 · Cited by 53
“W ITH the development of deep learning, text-to-speech (TTS) [1], [2] and voice conversion (VC) [3] technologies have made significant improvements.”
BackgroundWubet, Y. A., & Lian, K. (2022) · IEEE Access · doi:10.1109/access.2022.3200479 · Cited by 47
“Many state-of-the-art VC methods [23], [24], [25] have been proposed and implemented for parallel and non-parallel VC.”
MethodMany-to-Many Voice Transformer Network
Kameoka, H., Huang, W.-C., Tanaka, K., Kaneko, T., Hojo, N., & Toda, T. (2020) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2020.3047262 · Cited by 40
Self-citationGeneralized Spoofing Detection Inspired from Audio Generation Artifacts
Gao, Y., Vuong, T., Elyasi, M., Bharaj, G., & Singh, R. (2021) · Interspeech (USB) · doi:10.21437/interspeech.2021-1705 · Cited by 46
Many-To-Many Voice Conversion Using Conditional Cycle-Consistent Adversarial Networks
Lee, S., Ko, B., Lee, K., Yoo, I.-C., & Yook, D. (2020) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp40776.2020.9053726 · Cited by 46
“Among them, cycle-consistent adversarial network (CycleGAN) based approaches [10, 11] have shown the state-of-the-art sound quality of the converted speech without using any extra modules such as automatic speech Shindong Lee is now with Multimedia R&D Part, AI Lab., Kakao Enterprise. recognition…”
BackgroundMethodInfluentialGenerative Adversarial Networks in Human Emotion Synthesis: A Review
Hajarolasvadi, N., Ramírez, M. A., Beccaro, W., & Demirel, H. (2020) · IEEE Access · doi:10.1109/access.2020.3042328 · Cited by 42
“The drawback of WaveCycleGAN-VC is aliasing distortion that is avoided in WaveCycleGAN-VC2 by adding identity loss.”
MethodInfluentialKumar, L., & Singh, D. K. (2023) · Multimedia Tools and Applications · doi:10.1007/s11042-023-15138-x · Cited by 44
“Exemplar GAN (Ex-GAN) [21] produces a great result by utilizing the model picture data as a source of perspective pictures for eye-closing educational errands.”
RetractedDeng, J., Songli, W., Jian-mei, Y., Ji, L., & Wang, Y. (2024) · Pattern Recognition · doi:10.1016/j.patcog.2024.110692 · Cited by 33
Robust Disentangled Variational Speech Representation Learning for Zero-Shot Voice Conversion
Lian, J., Zhang, C., & Yu, D. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9747272 · Cited by 65
SpeakerGAN: Speaker identification with conditional generative adversarial network
Chen, L., Liu, Y., Xiao, W., Wang, Y., & Xie, H. (2020) · Neurocomputing · doi:10.1016/j.neucom.2020.08.040 · Cited by 34
An Adaptive-Learning-Based Generative Adversarial Network for One-to-One Voice Conversion
Dhar, S., Jana, N. D., & Das, S. (2022) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2022.3149858 · Cited by 40
“Seshadri et al. [25] incorporated Pulse Model in the Log-domain (PML) vocoder [26] as the first approach in the CycleGAN-VC framework for both speech feature extraction and speech reconstruction .”
BackgroundMethodResultInfluentialLuo, Z., Chen, J., Takiguchi, T., & Ariki, Y. (2019) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2019.2923951 · Cited by 35
“The MCD results are shown in Table II and Table III, GAN-1 and GAN-2 obtained slightly better results than the DBNs models, but the results were worse than those for the CycleGAN and dual-SANs.”
MethodInfluentialDualLip: A System for Joint Lip Reading and Generation
Chen, W., Tan, X., Xia, Y., Qin, T., Wang, Y., & Liu, T.-Y. (2020) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3394171.3413623 · Cited by 34
“er the task of lip generation as a dual task of lip reading and build a system named DualLip to jointly train two models for the two tasks by leveraging their task duality and unpaired/unlabeled data [14, 17, 31]: 1) Given unlabeled text, we first use a lip generation model to synthesize lip video, and then add the ps…”
BackgroundMethodChoi, Y., Jung, Y., & Kim, H. (2021) · IEEE Spoken Language Technology Workshop (SLT) · doi:10.1109/slt48900.2021.9383533 · Cited by 28
“Speech generation tasks such as text-to-speech and voice conversion have achieved great success in recent years with advances in deep learning [1–5].”
DeepConversion: Voice conversion with limited parallel training data
Zhang, M., Şişman, B., Zhao, L., & Li, H. (2020) · Speech Communication · doi:10.1016/j.specom.2020.05.004 · Cited by 27
Fu, C., Deng, Q., Shen, J., Mahzoon, H., & Ishiguro, H. (2022) · Sensors · doi:10.3390/s22030991 · Cited by 26
“Regarding the extraction of the F0 feature, we adopted the continuous wavelet transform (CWT) to decompose the F0 contour with 10 temporal scales as performed in previous research [45,47].”
MethodDuration Controllable Voice Conversion via Phoneme-Based Information Bottleneck
Lee, S.-H., Noh, H.-R., Nam, W.-J., & Lee, S. (2022) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2022.3156757 · Cited by 33
How Far Are We from Robust Voice Conversion: A Survey
Huang, T.-H., Lin, J.-H., & Lee, H.-Y. (2021) · IEEE Spoken Language Technology Workshop (SLT) · doi:10.1109/slt48900.2021.9383498 · Cited by 39
“Recently, several models were proposed to utilize non-parallel data [1, 2, 3, 4, 5].”
BackgroundTriAAN-VC: Triple Adaptive Attention Normalization for Any-to-Any Voice Conversion
Park, H. J., Yang, S. W., Kim, J. S., Shin, W., & Han, S. W. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10096642 · Cited by 40
Prananta, L., Halpern, B. M., Feng, S., & Scharenborg, O. (2022) · Interspeech (USB) · doi:10.21437/interspeech.2022-190 · Cited by 25
Deepfakes in digital media forensics: Generation, AI-based detection and challenges
Bendiab, G., Haiouni, H., Moulas, I., & Shiaeles, S. N. (2024) · Journal of Information Security and Applications · doi:10.1016/j.jisa.2024.103935 · Cited by 35
Zhu, Q., Xu, T., Xu, Y., & He, Y. (2021) · Industrial & Engineering Chemistry Research · doi:10.1021/acs.iecr.1c03197 · Cited by 24
VoiceGrad: Non-Parallel Any-to-Many Voice Conversion With Annealed Langevin Dynamics
Kameoka, H., Kaneko, T., Tanaka, K., Hojo, N., & Seki, S. (2024) · IEEE/ACM Transactions on Audio Speech and Language Processing · doi:10.1109/taslp.2024.3379901 · Cited by 32
“We previously reported a non-parallel VC method [20], [21] using a GAN variant called cycle-consistent GAN (CycleGAN) [22]–[24], which was originally proposed as a method for translating images using unpaired training examples.”
MethodSelf-citationA Comparative Study of Self-Supervised Speech Representation Based Voice Conversion
Huang, W.-C., Yang, S.-W., Hayashi, T., & Toda, T. (2022) · IEEE Journal of Selected Topics in Signal Processing · doi:10.1109/jstsp.2022.3193761 · Cited by 32
“To avoid the costly parallel data collection process, CycleGAN-based VC [13] was proposed to find the mapping function without explicit alignment using adversarial learning.”
Analysis of gender and identity issues in depression detection on de-identified speech
Lopez‐Otero, P., & Docío-Fernández, L. (2020) · Computer Speech & Language · doi:10.1016/j.csl.2020.101118 · Cited by 25
CycleGAN-based Non-parallel Speech Enhancement with an Adaptive Attention-in-attention MechanismPreprint
Yu, G., Wang, Y., Zheng, C., Wang, H., & Zhang, Q. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2107.13143 · Cited by 23
“To resolve this problem, CycleGAN was adopted for both standard parallel and non-parallel training in the SE area [14], [15], [16], which was originally proposed for unpaired image-to-image translation [17] and began to thrive in speech applications in recent years [18].”
MethodMulti-Task Learning Improves Synthetic Speech Detection
Mo, Y., & Wang, S. (2022) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp43922.2022.9746059 · Cited by 30
Style Equalization: Unsupervised Learning of Controllable Generative Sequence ModelsPreprint
Chang, J.-H. R., Shrivastava, A., Koppula, H. S., Zhang, X., & Tuzel, O. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.02891 · Cited by 24
“Instead of using speaker labels as input, Kameoka et al. (2018); Kaneko & Kameoka (2018); Kaneko et al. (2019a;b) group training samples by their speaker labels and apply adversarial learning to learn voice conversion models that change speaker voices while keeping the content of the input.”
He, Y., Seng, K. P., & Ang, L.-M. (2025) · IEEE Transactions on Neural Systems and Rehabilitation Engineering · doi:10.1109/tnsre.2025.3570383 · Cited by 22
“3) Improved CycleGAN-Based Conversion Model: In this work, we employ a network architecture inspired by CycleGAN-VC [31].”
Ghosh, S., Sarkar, S., Ghosh, S., Zalkow, F., & Jana, N. D. (2024) · Applied Intelligence · doi:10.1007/s10489-024-05380-7 · Cited by 23
“Furthermore, cycleGAN-based methods introduced the concept of unpaired VC, enabling conversion between speakers without parallel training data [25].”
Zhang, J., Liu, L., Chen, Y., Hu, Y.-J., Jiang, Y., Ling, Z.-H., & Dai, L.-R. (2020) · doi:10.21437/vcc_bc.2020-16 · Cited by 20
Atkar, G. B., & Jayaraju, P. (2021) · Neural Computing and Applications · doi:10.1007/s00521-021-05695-3 · Cited by 25
“Researchers proposed a cycle consistent adversarial network (CycleGAN) for nonparallel data-based VC training [10].”
BackgroundNoise-robust voice conversion with domain adversarial training
Du, H., Xie, L., & Li, H. (2022) · Neural Networks · doi:10.1016/j.neunet.2022.01.003 · Cited by 24
“Neural vocoder Parallel WaveGAN [56] was used to synthesize the converted speech.”
MethodInfluentialChu, M., Yang, M., Xu, C., Ma, Y., Wang, J., Fan, Z., Tao, Z., & Wu, D. (2023) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2023.3239551 · Cited by 26
Multimodal dance style transfer
Yin, W., Yin, H., Baraka, K., Kragić, D., & Björkman, M. (2023) · Machine Vision and Applications · doi:10.1007/s00138-023-01399-x · Cited by 20
GAZEV: GAN-Based Zero-Shot Voice Conversion Over Non-Parallel Speech Corpus
Zhang, Z., He, B., & Zhang, Z. (2020) · Interspeech (USB) · doi:10.21437/interspeech.2020-1710 · Cited by 20
Yu, G., Wang, Y., Wang, H., Zhang, Q., & Zheng, C. (2021) · Speech Communication · doi:10.1016/j.specom.2021.09.001 · Cited by 19
TGAVC: Improving Autoencoder Voice Conversion with Text-Guided and Adversarial Training
Tang, H., Zhang, X., Wang, J., Cheng, N., Zeng, Z., Xiao, E., & Xiao, J. (2021) · IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU) · doi:10.1109/asru51503.2021.9688088 · Cited by 25
“This paper is supported by National Key Research and Development Program of China under grant No. 2018YFB0204403, No. 2017YFB1401202 and No. 2018YFB1003500. voice conversion is very similar to that of image style transfer in computer vision, many generative adversarial networks (GAN) that have been successfully applie…”
BackgroundMethodInfluentialUnsupervised Learning of Disentangled Speech Content and Style Representation
Tjandra, A., Pang, R., Zhang, Y., & Karita, S. (2021) · Interspeech (USB) · doi:10.21437/interspeech.2021-1936 · Cited by 26
Expressive Voice Conversion: A Joint Framework for Speaker Identity and Emotional Style Transfer
Du, Z., Şişman, B., Zhou, K., & Li, H. (2021) · IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU) · doi:10.1109/asru51503.2021.9687906 · Cited by 31
Lakshmi, L., Kalyani, A. N., Madhuri, D. K., Potluri, S., Pandey, G. S., Ali, S., Khan, M. I., Awwad, F. A. A., & Ismail, E. A. A. (2023) · IEEE Access · doi:10.1109/access.2023.3337430 · Cited by 21
Guo, H., Liu, C., Ishi, C. T., & Ishiguro, H. (2023) · IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU) · doi:10.1109/asru57964.2023.10389651 · Cited by 26
“Voice conversion (VC) is a task that aims to modify a speaker’s voice characteristics, such as speaker identity[1], emotion[2], and accent[3] while preserving the linguistic content.”
BackgroundWang, Y., Yu, G., Wang, J., Wang, H., & Zhang, Q. (2020) · IEEE Access · doi:10.1109/access.2020.3029417 · Cited by 18
“studies on CycleGAN-based approaches for speech processing [22]–[24], we propose a relativistic-loss cycleconsistent GAN with multi-attention and dilated residual network (DRN) for single-channel speech enhancement.”
BackgroundMethodInfluentialGu, Y., Zhang, Z., Yi, X., & Zhao, X. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.02500 · Cited by 21
“The GAN-based models, e.g., CycleGAN-VC[7, 8, 9], StarGAN-VC[10, 11], are designed for a predefined set of speakers, and due to lack of explicit speaker embeddings, can hardly realize A2A VC. Learning from mel-spectrograms of source speeches, traditional pitch-edited speeches and converted speeches, we consider conten…”
MethodFace-based Voice Conversion: Learning the Voice behind a Face
Lu, H.-H., Weng, S.-E., Yen, Y.-F., Shuai, H.-H., & Cheng, W.-H. (2021) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3474085.3475198 · Cited by 23
“As parallel training data are more difficult to collect, a recent line of studies aims to train VC with non-parallel data, which can be mainly categorized into Generative Adversarial Network (GAN)-based [21, 23–25] and Autoencoder (AE)-based [7, 40, 41, 54, 55] approaches.”
BackgroundKumar, V., Kumar, A., & Shahnawazuddin, S. (2021) · Circuits Systems and Signal Processing · doi:10.1007/s00034-021-01885-5 · Cited by 17
Fu, C., Liu, C., Ishi, C. T., & Ishiguro, H. (2022) · Speech Communication · doi:10.1016/j.specom.2022.09.002 · Cited by 17
Cycle-consistent Adversarial Networks for Non-parallel Vocal Effort Based Speaking Style Conversion
Seshadri, S., Juvela, L., Yamagishi, J., Räsänen, O. J., & Alku, P. (2019) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2019.8682648 · Cited by 28
“The training was run for 100 epochs with the identity mapping loss, Lid being dropped after 50 epochs (similar to [22]).”
MethodCVC: Contrastive Learning for Non-Parallel Voice Conversion
Li, T., Liu, Y., Hu, C., & Zhao, H. (2021) · Interspeech (USB) · doi:10.21437/interspeech.2021-137 · Cited by 18
Wani, T. M., Qadri, S. A. A., Wani, F. A., & Amerini, I. (2024) · Foundations and Trends® in Privacy and Security · doi:10.1561/3300000048 · Cited by 16
Wang, M., Xiong, Z., Zhao, J., Zhou, S., Wang, Y., Ikram, R. M. A., Wang, L., & Tan, S. K. (2025) · Land · doi:10.3390/land14030578 · Cited by 16
Raman spectrum model transfer method based on Cycle-GAN
Wang, Z., Yang, Z., Song, X., Zhang, H., Sun, B., Zhai, J., Yang, S., Xie, Y., & Liang, P. (2023) · Spectrochimica Acta Part A Molecular and Biomolecular Spectroscopy · doi:10.1016/j.saa.2023.123416 · Cited by 16
Voice Conversion with Cyclic Recurrent Neural Network and Fine-tuned Wavenet Vocoder
Tobing, P. L., Wu, Y.-C., Hayashi, T., Kobayashi, K., & Toda, T. (2019) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2019.8682156 · Cited by 22
“In this paper, inspired by the CycleGAN architecture [5], we present a novel approach to improving a parallel VC system, especially for a WaveNet fine-tuning framework [15], by”
BackgroundMethodMany-to-Many Voice Conversion Using Cycle-Consistent Variational Autoencoder with Multiple Decoders
Yook, D., Leem, S.-G., Lee, K., & Yoo, I.-C. (2020) · The Speaker and Language Recognition Workshop (Odyssey) · doi:10.21437/odyssey.2020-31 · Cited by 20
Enhancing Voice Cloning Quality through Data Selection and Alignment-Based Metrics
González-Docasal, A., & Álvarez, A. (2023) · Applied Sciences · doi:10.3390/app13148049 · Cited by 18
PMVC: Data Augmentation-Based Prosody Modeling for Expressive Voice Conversion
Deng, Y., Tang, H., Zhang, X., Wang, J., Cheng, N., & Xiao, J. Z. (2023) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3581783.3613800 · Cited by 18
“GAN-based methods [10, 11] are employed to perform VC task with advantages of high efficiency and diversity.”
MethodTowards Fine-Grained Prosody Control for Voice Conversion
Lian, Z., Zhong, R., Wen, Z., Liu, B., & Tao, J. (2021) · International Symposium on Chinese Spoken Language Processing (ISCSLP) · doi:10.1109/iscslp49672.2021.9362110 · Cited by 15
“To address this problem, researches have proposed some methods for a voice conversion system that does not require the parallel training data [11, 12, 13].”
MethodZero-Shot Voice Conversion with Adjusted Speaker Embeddings and Simple Acoustic Features
Tan, Z., Wei, J., Xu, J., He, Y., & Lu, W. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9414975 · Cited by 20
“Many methods based on non-parallel data have been proposed, such as VC systems using Generative adversarial network (GAN) [14,15,16], which have been proved to be able to perform many-to-many VC with non-parallel data.”
MethodImproving the Efficiency of Dysarthria Voice Conversion System Based on Data Augmentation
Zheng, W.-Z., Han, J.-Y., Chen, C.-Y., Chang, Y.-J., & Lai, Y.-H. (2023) · IEEE Transactions on Neural Systems and Rehabilitation Engineering · doi:10.1109/tnsre.2023.3331524 · Cited by 15
“Next, the synthesized normal speech n ( t ) is used to generate the corresponding paired dysarthric corpus s ( t ) through the StarGAN-VC [29], [30] model.”
MethodCyclegean: Cycle Generative Enhanced Adversarial Network for Voice Conversion
Zhang, X., Wang, J., Cheng, N., Xiao, E., & Xiao, J. Z. (2021) · IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU) · doi:10.1109/asru51503.2021.9687948 · Cited by 18
“Kaneko et al. [28]propose to use Cycle-consistent Generative Adversarial Networks (CycleGAN) in the computer vision task.”
MethodBargum, A. R., Serafin, S., & Erkut, C. (2024) · Frontiers in Signal Processing · doi:10.3389/frsip.2024.1339159 · Cited by 20
Akhter, M. T., Banerjee, P., Dhar, S., Ghosh, S., & Jana, N. D. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-48309-7_20 · Cited by 15
GLGAN-VC: A Guided Loss-Based Generative Adversarial Network for Many-to-Many Voice Conversion
Dhar, S., Jana, N. D., & Das, S. (2023) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2023.3335119 · Cited by 15
“D t-distributed stochastic neighbor embedding (t-SNE) visualization [44], and mel-cepstral coefficients (MCEP) trajectory plot [11].”
Conan's Bow Tie: A Streaming Voice Conversion for Real-Time VTuber Livestreaming
Chen, Q., Gu, Z. P., Lü, L., Xu, X., Ba, Z., Lin, F., Liu, Z., & Ren, K. (2024) · ACM International Conference on Intelligent User Interfaces (IUI) · doi:10.1145/3640543.3645146 · Cited by 15
“…their extensive utilization in VC methodologies, encompassing various models such as Deep Neural Networks (DNN)[12, 33, 35, 51], Long-Short Term Memory (LSTM)[32, 36, 46], Transformers[21, 47, 53, 54], Auto-encoders[40], Generative Ad-versarial Networks (GAN)[19, 22, 37], and diffusion models[39].”
BackgroundMethodFault detection in new wind turbines with limited data by generative transfer learning
Jonas, S., & Meyer, A. (2025) · Energy and AI · doi:10.1016/j.egyai.2025.100626 · Cited by 13
“Domain mapping with CycleGAN has been used in various applications, e.g. in medical settings [26], voice conversion [27], machine fault diagnosis [20], or also for synthetic data augmentation in WT icing detection [28].”
Generative adversarial networks: Foundations and applications
Kaneko, T. (2018) · Nippon Onkyo Gakkaishi/Acoustical science and technology/Nihon Onkyo Gakkaishi · doi:10.1250/ast.39.189 · Cited by 14
Self-citationCycleGAN-VC-GP: Improved CycleGAN-based Non-parallel Voice Conversion
Wang, C., & Yu, Y. (2020) · IEEE International Conference on Communication Technology (ICCT) · doi:10.1109/icct50939.2020.9295938 · Cited by 13
“Recently, with the rapid development of GAN, CycleGAN has been used in non-parallel voice conversion [14-15].”
Method
Metadata
- Authors
- Kaneko, T., & Kameoka, H.
- Affiliations
- NTT (Japan) (JP)
- Venue
- 2018 26th European Signal Processing Conference (EUSIPCO)
- Publisher
- IEEE
- Published
- 2018-09
- Pages
- 2100-2104
- OpenAlex
- W2902070858
- Cited by
- 358
- References
- 61
- Citations by year
- 2026: 19 · 2025: 36 · 2024: 50 · 2023: 66 · 2022: 58 · 2021: 63 · 2020: 47 · 2019: 18 · 2018: 1
- FWCI
- 13.16
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- European Signal Processing Conference (EUSIPCO) (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Speech Recognition and Synthesis
- Physical Sciences › Computer Science › Signal Processing › Speech and Audio Processing
- Health Sciences › Health Professions › Speech and Hearing › Voice and Speech Disorders
- Keywords
- voice conversion, non-parallel voice conversion, CycleGAN, gated convolutional neural network, pseudo data generation, over-smoothing, modulation spectra
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.