The DOI matches its Crossref record; 32 of the 32 references it cites were found in a registry or an index; 2 of the 22 sources its claims cite were checked for support.
A Lip Sync Expert Is All You Need for Speech to Lip Generation In the Wild
Open accessAbstract
In this work, we investigate the problem of lip-syncing a talking face video of an arbitrary identity to match a target speech segment. Current works excel at producing accurate lip movements on a static image or videos of specific people seen during the training phase. However, they fail to accurately morph the lip movements of arbitrary identities in dynamic, unconstrained talking face videos, resulting in significant parts of the video being out-of-sync with the new audio. We identify key reasons pertaining to this and hence resolve them by learning from a powerful lip-sync discriminator. Next, we propose new, rigorous evaluation benchmarks and metrics to accurately measure lip synchronization in unconstrained videos. Extensive quantitative evaluations on our challenging benchmarks show that the lip-sync accuracy of the videos generated by our Wav2Lip model is almost as good as real synced videos. We provide a demo video clearly showing the substantial impact of our Wav2Lip model, and also publicly release the code, models, and evaluation benchmarks on our website.
Overview
CoChat Summary
TL;DR
Wav2Lip lip-syncs arbitrary talking-face videos in the wild by training a generator against a pretrained, accurate lip-sync expert discriminator rather than a weak GAN discriminator. It reports lip-sync accuracy almost as good as real synced videos and introduces new benchmarks and metrics for unconstrained evaluation.
Key findings
- On LRS2, Wav2Lip reaches LSE-D 6.386 and LSE-C 7.789, close to real videos (6.736, 7.838) and much better than LipGAN (10.33, 3.199) and Speech2Vid (14.23, 1.587).
- The expert lip-sync discriminator is 91% accurate on the LRS2 test set versus 56% for LipGAN’s discriminator; in ablation, the frozen T_v=5 expert reaches 91.6% off-sync accuracy with LSE-D 6.386/LSE-C 7.789, while fine-tuning it drops to 73.6% and LSE-D 9.953/LSE-C 3.508.
- On ReSyncED, Wav2Lip models are preferred over existing methods or unsynced videos more than 90% of the time; adding the visual quality discriminator improves FID but causes a minor drop in sync accuracy.
- The paper claims Wav2Lip is the first speaker-independent model to generate videos with lip-sync accuracy matching real synced videos.
Methods
- Generator follows a LipGAN-like architecture with identity encoder, speech encoder, and face decoder; trained with L1 reconstruction loss, expert sync loss from a frozen pretrained lip-sync discriminator, and adversarial loss from a visual quality discriminator, with sync weight 0.03 and adversarial weight 0.07.
- The expert discriminator modifies SyncNet using color images, deeper residual skip connections, and cosine-similarity with binary cross-entropy; trained on LRS2 train split (~29 hours), batch size 64, T_v=5, Adam with initial learning rate 1e-3.
- Evaluation uses new LSE-D/LSE-C metrics from public SyncNet, benchmarks built from LRS2 (14K pairs), LRW (28K pairs), and LRS3 (14K pairs), plus ReSyncED real-world YouTube videos in Dubbed, Random, and TTS categories, with human evaluation by 14 evaluators.
Limitations
- Stated in the paper: human evaluations indicate scope for improvement when lip-syncing TTS-generated speech, and using the visual quality discriminator causes a minor drop in sync accuracy, creating a trade-off between visual quality and sync accuracy.
- Evident from the text: the model is trained only on the LRS2 train set, and the real-world benchmark is collected from YouTube with manually selected video types and synthetic translated speech.
- Evident from the text: the human evaluation uses 14 evaluators, and the automatic lip-sync metrics rely on a public SyncNet model trained on a different, non-public dataset.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
2 Valid · 2 Not valid · 13 Not checked · 59 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“We find that these models do not adequately penalize wrong lip shapes, either as a result of using only reconstruction losses or weak lip-sync discriminators.”
- Also stated in: § 3 ACCURATE SPEECH-DRIVEN LIP-SYNCING FOR VIDEOS IN THE WILD · plain
“We argue that the loss functions, namely the L1 reconstruction loss used in both the existing works [17, 18] and the discriminator loss in LipGAN [18] are inadequate to penalize inaccurate lip-sync generation.”
What the sources say
- Partly supportedcites [17] [17] Amir Jamaludin, Joon Son Chung, and Andrew Zisserman. 2019. You said that?: Synthesising talking faces from audio. International Journal of Computer Vision 127, 11-12 (2019), 1767-1779. - sourceExperimental — accuracy not yet measured
Section 3.2, “Loss Function” (page unavailable), says the model is trained with image-space and content losses, supporting the reconstruction-loss part of the description. The “Correctness of the Generated Lip Shape” section evaluates lip synchronisation, but does not conclude that these losses inadequately penalise wrong lip shapes or discuss weak lip-sync discriminators. (LLM, conf=0.78)
- Not checked yetcites [18] [18] Prajwal KR, Rudrabha Mukhopadhyay, Jerin Philip, Abhishek Jha, Vinay Namboodiri, and CV Jawahar. 2019. Towards Automatic Face-to-Face Translation. In Proceedings of the 27th ACM International Conference on Multimedia . ACM, 1428-1436. -
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § 3 ACCURATE SPEECH-DRIVEN LIP-SYNCING FOR VIDEOS IN THE WILD · plain
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Thus, we believe that our model can enable a wide range of real-world applications where previous speaker-independent lipsyncing approaches [17, 18] struggle to produce satisfactory results.”
What the sources say
- Partly supportedcites [17] [17] Amir Jamaludin, Joon Son Chung, and Andrew Zisserman. 2019. You said that?: Synthesising talking faces from audio. International Journal of Computer Vision 127, 11-12 (2019), 1767-1779. - sourceExperimental — accuracy not yet measured
Our copy supports the identification part: Speech2Vid is a prior speaker-independent lip-syncing method — the Introduction states it produces talking-face video 'even when using images and audio outside of the training dataset', Sect. 2 contrasts methods 'only trained for one identity' with those 'applicable to any at test time', and Sect. 7 (Summary) concludes it 'is able to generate videos of any identity speaking from any source of input audio'. However, the paper nowhere says it struggles to produce satisfactory results; on the contrary, 'Correctness of the Generated Lip Shape' reports sync retrieval accuracy well above chance (79.7–83.9% vs 9.7% chance, against 89.5% on ground truth), and the re-dubbing 'Results and Discussion' calls the CNN blending 'very effective'. The unsupported part of the claim is the negative quality judgement; the closest the paper comes is acknowledging limitations — blur ('blur seems to be a good tell that an image is generated', Sect. 3.3 / Identity Preservation) and 'the difficulty in making the chins move with the mouth' (Results and Discussion, Sect. 6). (LLM, conf=0.72)
- Not checked yetcites [18] [18] Prajwal KR, Rudrabha Mukhopadhyay, Jerin Philip, Abhishek Jha, Vinay Namboodiri, and CV Jawahar. 2019. Towards Automatic Face-to-Face Translation. In Proceedings of the 27th ACM International Conference on Multimedia . ACM, 1428-1436. -
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our method outperforms previous approaches by a large margin indicating the significant effect of strong lip-sync discrimination.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Using the LRW test set, we create 28 K pairs”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, we observe a minor drop in sync accuracy after using the visual quality discriminator.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
54 more claims
Create a free account to see all 76 claims and to verify the 20 references still open.
- It may state two separate claims as one. 10
- Probably not a substantial claim. 9
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?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“These methods, though improved generalization, however, concentrate on partial facial attributes, such as lip movements or emotional biases, which are easily countered by specific countermeasures [44, 45].”
From Attention-based Multimodal learning framework for Generalized Audio-Visual Deepfake Detection (2023) · cites this paper as [44]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“pioneer talking-head generation works typically utilize biometric signals such as facial keypoints [22, 28, 30, 45], or 3D priors [11, 13, 16, 23, 32, 43, 44] as intermediate motion representation to animate the reference face while ensuring lip synchronization.”
From MoCha: Towards Movie-Grade Talking Character Synthesis (2025) · cites this paper as [28]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Since full-body shots and background environments are included in our videos, Wav2Lip [19] is used to preserve the additional context in the video.”
From CLFormer: a cross-lingual transformer framework for temporal forgery localization (2025) · cites this paper as [19]
The cited source says what the claim attributes to it.
In the citing paper
“The output quality of the visual generator Wav2Lip [46] and audio generator SV2TTS [27] used for generating LAV-DF is not sufficient for state-of-the-art detection methods.”
From AV-Deepfake1M: A Large-Scale LLM-Driven Audio-Visual Deepfake Dataset (2024) · cites this paper as [46]
The cited source says what the claim attributes to it.
In the citing paper
“Many deep learning driven technologies [39, 105, 106, 107] have been introduced to address the alignment of AV signals in a self-supervised way, i . e ., automatically creating positive and negative A-V sync/out-of-sync pairs.”
From Deep learning for visual speech analysis: A survey (2024) · cites this paper as [39]
2 more claims about this paper
Create a free account to see every claim other papers make about this one.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 4.4 Real-World Evaluation
“| Without Lip-syncing | | 16.89 | 2.557 | - | 0.11 | 4.67 | 3.32 | 8.32% |”
§ 4.4 Real-World Evaluation
“| LipGAN [18] | TTS | 10.90 | 3.279 | 11.91 | 2.87 | 3.69 | 3.14 | 1.64% | | Wav2Lip (ours) | | 6.659 | 8.126 | 12.77 | 3.98 | 3.87 | 3.92 | 41.2% | | Wav2Lip + GAN (ours) | | 7.225 | 7.651 | 11.15 | 3.85 | 4.13 | 4.05 | 51.2% |”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 5 APPLICATIONS & FAIR USE
“We can also lip-sync dubbed movies making them pleasant to watch (Table 2, first block).”
§ 4.4 Real-World Evaluation
“| Unsynced Orig. Videos | Random | 12.63 | 0.896 | - | 0.21 | 4.81 | 3.07 | 3.15% |”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 4.4 Real-World Evaluation
“| Unsynced Orig. Videos | Random | 12.63 | 0.896 | - | 0.21 | 4.81 | 3.07 | 3.15% |”
§ 4.4 Real-World Evaluation
“The first three parameters are scored between 1 -5, and (d) is a single-choice voting, and we report the percentage of votes obtained by a model.”
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
Image quality assessment: from error visibility to structural similarity
Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004) · IEEE Transactions on Image Processing · doi:10.1109/tip.2003.819861 · Cited by 57,892
(2010) · Cited by 8,261
Instance Normalization: The Missing Ingredient for Fast Stylization
Ulyanov, D., Vedaldi, A., & Lempitsky, V. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1607.08022 · Cited by 2,775
International Journal of Computer Vision
(2013) · International Journal of Computer Vision · doi:10.1007/11263.1573-1405 · Cited by 2,340
Deepfakes and beyond: A Survey of face manipulation and fake detection
Tolosana, R., Vera-Rodriguez, R., Fierrez, J., Morales, A., & Ortega-Garcia, J. (2020) · Information Fusion · doi:10.1016/j.inffus.2020.06.014 · Cited by 1,237
DeepFakes and Beyond: A Survey of Face Manipulation and Fake Detection
Tolosana, R., Vera-Rodríguez, R., Fierrez, J., Morales, A., & Ortega-Garcia, J. (2020) · Information Fusion · doi:10.48550/arxiv.2001.00179 · Cited by 1,234
An audio-visual corpus for speech perception and automatic speech recognition
Cooke, M., Barker, J., Cunningham, S., & Shao, X. (2006) · The Journal of the Acoustical Society of America · doi:10.1121/1.2229005 · Cited by 1,144
Synthesizing Obama: learning lip sync from audio
Suwajanakorn, S., Seitz, S. M., & Kemelmacher-Shlizerman, I. (2017) · ACM Transactions on Graphics · doi:10.1145/3072959.3073640 · Cited by 1,057
Deep Audio-Visual Speech Recognition
Afouras, T., Chung, J. S., Senior, A., Vinyals, O., & Zisserman, A. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2018.2889052 · Cited by 801
Research works printed but not linked: 3
- Prajwal KR , Rudrabha Mukhopadhyay , Jerin Philip , Abhishek Jha , Vinay Namboodiri , and CV Jawahar . 2019 . Towards Automatic Face-to-Face Translation . In Proceedings of the 27th ACM International Conference on Multimedia. ACM, 1428--1436 . Prajwal KR, Rudrabha Mukhopadhyay, Jerin Philip, Abhishek Jha, Vinay Namboodiri, and CV Jawahar. 2019. Towards Automatic Face-to-Face Translation. In Proceedings of the 27th ACM International Conference on Multimedia. ACM, 1428--1436.
- Andrew L Maas , Awni Y Hannun , and Andrew Y Ng . 2013 . Rectifier nonlinearities improve neural network acoustic models . In Proc. icml , Vol. 30 . 3. Andrew L Maas, Awni Y Hannun, and Andrew Y Ng. 2013. Rectifier nonlinearities improve neural network acoustic models. In Proc. icml, Vol. 30. 3.
- NPD. 2016. 52 Percent of Millennial Smartphone Owners Use their Device for Video Calling According to The NPD Group. https://www.npd.com/wps/portal/npd/us/news/press-releases/2016/52-percent-of-millennial-smartphone-owners-use-their-device-for-video-calling-according-to-the-npd-group/ NPD. 2016. 52 Percent of Millennial Smartphone Owners Use their Device for Video Calling According to The NPD Group. https://www.npd.com/wps/portal/npd/us/news/press-releases/2016/52-percent-of-millennial-smartphone-owners-use-their-device-for-video-calling-according-to-the-npd-group/
Cited by
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
Generative artificial intelligence: a systematic review and applications
Sengar, S. S., Hasan, A. B., Kumar, S., & Carroll, F. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-20016-1 · Cited by 467
AI-generated characters for supporting personalized learning and well-being
Pataranutaporn, P., Danry, V., Leong, J., Punpongsanon, P., Novy, D., Maes, P., & Sra, M. (2021) · Nature Machine Intelligence · doi:10.1038/s42256-021-00417-9 · Cited by 358
Pose-Controllable Talking Face Generation by Implicitly Modularized Audio-Visual Representation
Zhou, H., Sun, Y., Wu, W., Loy, C. C., Wang, X., & Liu, Z. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr46437.2021.00416 · Cited by 370
Flow-guided One-shot Talking Face Generation with a High-resolution Audio-visual Dataset
Zhang, Z., Li, L., Ding, Y., & Fan, C. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr46437.2021.00366 · Cited by 355
Metadata
- Authors
- Prajwal, K. R., Mukhopadhyay, R., Namboodiri, V. P., & Jawahar, C. V.
- Affiliations
- International Institute of Information Technology, Hyderabad (IN)
- University of Bath (GB)
- Venue
- Proceedings of the 28th ACM International Conference on Multimedia
- Publisher
- ACM
- Published
- 2020-10-12
- Pages
- 484-492
- OpenAlex
- W3081492798
- Cited by
- 897
- References
- 31
- Citations by year
- 2026: 157 · 2025: 235 · 2024: 211 · 2023: 166 · 2022: 75 · 2021: 50 · 2020: 3
- FWCI
- 45.95
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- ACM International Conference on Multimedia (ACM MM) (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Pure (University of Bath) (submittedVersion, other-oa)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Signal Processing › Speech and Audio Processing
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Generative Adversarial Networks and Image Synthesis
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Face recognition and analysis
- Keywords
- lip synchronization, unconstrained video, evaluation benchmark, audio-video synchronization
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-29, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-09-29 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: A Lip Sync Expert Is All You Need for Speech to Lip Generation In the Wild. https://app.cochat.ai/papers/10.1145/3394171.3413532
- The paper
- Prajwal, K. R., Mukhopadhyay, R., Namboodiri, V. P., & Jawahar, C. V. (2020). A Lip Sync Expert Is All You Need for Speech to Lip Generation In the Wild. Proceedings of the 28th ACM International Conference on Multimedia. https://doi.org/10.1145/3394171.3413532
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.