Learn2Chat: Rethinking Dyadic Talking Heads via Interaction-Modulated Monologic Priors

Cheng Xu; Haoxin Yang; Shengfeng He; Siyue Chen; Xuemiao Xu; Yihong Lin; Zikai Huang

Dyadic talking-head motion is interaction modulation of frozen monologic speech-motion priors, not a new joint generator.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-14 12:41 UTC pith:6XUQBOGJ

load-bearing objection Solid modular reformulation of dyadic head motion as modulation over frozen monologic priors; DualTalk SOTA is real, but the “clean interaction residual” story is only partly proven. the 4 major comments →

arxiv 2607.10313 v1 pith:6XUQBOGJ submitted 2026-07-11 cs.GR cs.SD

Learn2Chat: Rethinking Dyadic Talking Heads via Interaction-Modulated Monologic Priors

Zikai Huang , Siyue Chen , Xuemiao Xu , Haoxin Yang , Cheng Xu , Yihong Lin , Shengfeng He This is my paper

classification cs.GR cs.SD

keywords dyadic motion generationaudio-driven facial animationmonologic priorsinteraction modulationmotion factorizationtalking headsdigital humans

verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Most systems that animate two people in conversation either need the partner's ground-truth motion or train a single model that mixes speech-driven facial motion with social feedback. This paper argues that the speech-to-motion map is already well learned by monologic talking-head models, so the real missing piece is only the residual social adjustment. Learn2Chat freezes a pretrained monologic generator, factors dyadic sequences into speech semantics plus an interaction latent, and then predicts that latent from the two audio streams alone. The predicted latent modulates the monologic base motion through the decoder, producing coordinated head motion without relearning lip sync from scratch. On the DualTalk benchmark the method leads quantitative metrics and user preference, and the same interaction module works when the monologic backbone is swapped at test time.

Core claim

Dyadic conversational head motion is better generated by modulating a frozen, pretrained monologic speech-motion prior with an audio-predicted interaction latent than by training a unified dyadic generator that entangles self-speech and partner feedback. The Monologic-Anchored Motion Factorization isolates a clean interaction residual; Cross-Attentive Interaction Latent Prediction recovers that residual from paired speech; the residual then steers the monologic base via AdaLN, yielding higher fidelity, better inter-speaker coordination, and backbone-agnostic reuse.

What carries the argument

Monologic-Anchored Motion Factorization: a shared semantics encoder plus an asymmetric interaction encoder that maps monologic motion to a learnable neutral prior and dyadic motion to a stochastic residual; swap and cycle losses force the residual to carry only the domain gap. That residual is later predicted from dual audio by cross-attentive streams and injected through AdaLN into the frozen monologic decoder.

Load-bearing premise

That monologic head motion is an interaction-neutral baseline whose residual after factorization is a pure, audio-predictable social modulation that the monologic manifold can cleanly separate from speech content.

What would settle it

If swapping the monologic backbone at test time (or ablating the factorization stage) collapses inter-speaker correlation and Fréchet distances back to monologic or non-factorized levels on DualTalk, the claim that the residual is a transferable interaction latent fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

Solid modular reformulation of dyadic head motion as modulation over frozen monologic priors; DualTalk SOTA is real, but the “clean interaction residual” story is only partly proven.

read the letter

The useful takeaway is simple: freeze a good monologic speech-to-motion model, factor dyadic motion into shared semantics plus a stochastic interaction latent, then predict that latent from paired audio with dual-stream cross-attention and AdaLN modulation. On DualTalk that beats monologic baselines, adapted motion-conditioned methods, and UniLS on FD/P-FD/MSE/rPCC (test and OOD) with DEEPTalk and UniTalker, plus a cleaner user-study win on interaction naturalness. Parameter count is modest relative to DualTalk/UniLS, and the supplement shows cross-backbone swap and better listener-side numbers.

What is actually new is the engineering of the separation, not a new generative family. Shared ES on monologic and dyadic motion, learnable monologic “neutral” prior zm, swap + cycle losses, then frozen factorization while training only the audio-to-z path with recon + InfoNCE. That is a clearer prior-reuse story than UniLS’s joint latent or DualTalk’s partner-motion conditioning. Ablations (Tables IV–V) move in the right direction when factorization or dual-stream CA is removed, and the math is ordinary VAE/Transformer/AdaLN with no internal contradiction.

The stress-test lands partially. DualTalk audio is conversational, so Gmono(A) already carries turn-taking and partner-conditioned prosody; Lswap can let z absorb any residual that turns monologic-looking motion into dyadic motion, not only pure social feedback. There is no MI or content-swap probe that z is speech-independent. So the “clean interaction modulation” narrative is stronger than the evidence; gains could partly be residual fitting on this benchmark. Fairness of the adapted L2L/DIM/DualTalk baselines is also a bit soft (monologic motion as partner condition is not their native setting). Single-benchmark, head-only, no code/error bars—standard for the area, not fatal.

This is for people building conversational avatars who already have monologic backbones and want a lightweight interaction adapter. The central empirical claim holds; the interpretation is a bit oversold. I would send it to referees and would cite the DualTalk numbers and the modular recipe if I were shipping a similar system. Worth a reading-group slot if the group cares about talking heads.

Referee Report

4 major / 0 minor

Summary. Learn2Chat reformulates audio-driven dyadic 3D head-motion generation as interaction modulation over frozen pretrained monologic generators. A Monologic-Anchored Motion Factorization (shared semantics encoder ES, stochastic interaction encoder EI for dyadic motion, learnable monologic interaction prior zm, and AdaLN decoder) is trained with reconstruction, swap, cycle, and KL losses to obtain an interaction latent z. A dual-stream cross-attentive audio encoder then predicts za from paired Wav2Vec2 features, aligned to motion-derived zd via InfoNCE, and modulates monologic semantics at inference. On DualTalk (test and OOD), with DEEPTalk and UniTalker backbones, the method reports SOTA FD/P-FD/MSE/rPCC, favorable parameter count, a user study (n=25), ablations, listener-role analysis, and cross-backbone transfer.

Significance. If the empirical gains hold under fair audio-only deployment, the paper offers a practical and timely alternative to large jointly trained dyadic generators: reuse strong monologic speech–motion models and learn a comparatively lightweight interaction adapter. The dual-stream cross-attention design, two-stage freeze-then-predict protocol, DualTalk SOTA tables, user study, parameter comparison, and backbone-swap results in the supplement are concrete strengths for TVCG-style conversational avatar work. Even if the residual is not a pure social-interaction factor, a residual-modulation recipe that improves coordination metrics while remaining plug-and-play would still be useful to the community.

major comments (4)

Sec. III-A–B, Eqs. (1)–(10): the central narrative treats xm = Gmono(Ai) as an interaction-neutral canonical baseline and zd as clean partner-induced modulation. DualTalk streams are multi-round conversational audio (turn-taking, backchannels, partner-conditioned prosody), so Gmono(Ai) already sees interaction-correlated speech; Lswap then forces D(ES(xm), zd) ≈ xd, so zd can absorb monologic prediction error and speech-content residuals as well as social feedback. The manuscript never tests speech–interaction independence (e.g., content-swap, mutual information between z and phoneme/text features, or holding A fixed while swapping partner context). Please either add such probes or substantially soften claims of “clean interaction representations” / “separates intrinsic speech-driven motion from social interaction effects” throughout abstract, intro, and conclusion.
Sec. III-B factorization data construction is under-specified relative to the monologic-anchor claim. It is unclear whether xm used in Lrec/Lswap is (i) Gmono applied to DualTalk conversational audio only, (ii) real monologic corpus motion, or (iii) DualTalk segments treated as monologic. If (i), the “semantic manifold learned from monologic data” is largely the monologic model’s residual on dyadic speech, which weakens the prior-reuse story and the interpretation of Table V. Clarify the exact sources of xm/xd, whether Gmono is frozen DualTalk-retrained or original monologic pretraining, and report a control that freezes ES/D trained only on true monologic data if available.
Sec. IV-B baseline protocol: L2L, DIM, and DualTalk are evaluated audio-only by substituting partner GT motion with monologic base motions. That adaptation is deployment-relevant but systematically handicaps methods designed for GT partner motion; UniLS is the only native audio-only dyadic peer. Please report (or clearly mark as oracle) the original motion-conditioned DualTalk/L2L/DIM numbers with GT partner motion alongside the adapted setting, and avoid framing large gaps vs adapted DualTalk as pure evidence that joint dyadic learning is inferior. The UniLS comparison and parameter table (Table III) should carry more of the SOTA argument.
Table IV OOD row for Ours lists jaw MSE as 0.60 while Tables I–II and the same table’s other rows place jaw MSE near 1.4; this looks like a transcription error and affects ablation interpretation. Please audit all metric tables (including supplement listener/cross-backbone tables) for consistency of scaling (×10−k) and values, and re-state which checkpoint/split each ablation uses.

Circularity Check

0 steps flagged

No circularity: empirical factorization-and-modulation pipeline supervised against DualTalk ground truth; SOTA claims are held-out metrics, not inputs renamed as predictions.

full rationale

Learn2Chat is a standard two-stage ML system paper, not a first-principles derivation. Monologic bases come from a frozen pretrained G_mono (Eq. 1); factorization is trained with L_rec, L_swap, cycle, and KL against observed monologic/dyadic motions (Eqs. 8–13); interaction latents are then predicted from dual audio with L_motion + InfoNCE against motion-derived z_d (Eqs. 15–18). None of these quantities is defined as the evaluation target: FD/P-FD/MSE/rPCC and user-study scores are measured on held-out DualTalk test/OOD splits against external baselines. There is no self-definitional loop (z is a learned stochastic latent, not the claim), no fitted constant re-labeled as a prediction, no load-bearing uniqueness theorem from overlapping authors, and no ansatz smuggled in via self-citation. Concerns that DualTalk monologic bases may already carry conversational style, or that the residual may not be pure interaction, are validity/assumption issues about what the residual represents—not circular reductions of the reported metrics to their training inputs. The derivation chain is therefore self-contained and non-circular.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 2 invented entities

Load-bearing content is mostly standard ML/graphics machinery plus domain modeling choices. Free parameters are ordinary training hyperparameters and adaptive loss weights. Axioms are domain assumptions (FLAME, Wav2Vec2, monologic motion as interaction-neutral). Invented entities are the interaction latent and the monologic-anchored factorization scheme, which are architectural constructs evaluated only inside this pipeline, not independently measured physical quantities.

free parameters (3)

Adaptive loss weights ω_i for Lrec, Lswap, Lsem-cyc, Lint-cyc, LKL, LInfoNCE
Learnable per-term weights (supplement Eq. 1) that balance the multi-objective factorization and prediction losses; values are fit during training rather than fixed a priori.
Monologic interaction prior parameters μ_m, σ_m
Trainable Gaussian parameters defining the neutral-interaction state for monologic motion (Sec. III-B); fitted so that dyadic residual is attributed to EI.
Architecture and training schedule (latent dims, Transformer depth L, 3000+1000 epochs, lr 1e-4, batch 64, window 200)
Hand-chosen hyperparameters that control capacity and optimization; not derived from theory.

axioms (4)

domain assumption A pretrained monologic generator Gmono already encodes the speech-driven component of head motion sufficiently that dyadic residual can be treated as interaction modulation.
Stated in Sec. I and III-A as the reformulation premise; without it factorization does not isolate social feedback.
domain assumption FLAME parameters (expression, jaw, relative neck pose) plus Wav2Vec2 features are adequate representations for conversational head dynamics.
Sec. IV-A and supplement I-A; standard in the cited talking-head literature.
ad hoc to paper Layer-normed shared semantics encoder plus swap/cycle losses yield a domain-invariant speech-semantic manifold free of interaction leakage.
Sec. III-B design choice; supported by ablations but not independently proven.
ad hoc to paper Bidirectional conversational coupling is captured by symmetric dual-stream self- and cross-attention on paired audio CLS tokens.
Sec. III-C architectural assumption; single-stream ablation degrades rPCC but does not prove uniqueness.

invented entities (2)

Interaction latent z (zd from motion EI; za from dual-audio encoder) no independent evidence
purpose: Stochastic variable that modulates monologic semantics via AdaLN to produce dyadic motion.
Defined and trained only inside Learn2Chat; no external measurement (e.g., independent social-signal annotation) validates z outside reconstruction/alignment losses.
Monologic-Anchored Motion Factorization scheme (ES, EI, D with swap/cycle objectives) no independent evidence
purpose: Disentangle speech-driven semantics from partner-conditioned modulation using monologic data as anchor.
Paper-specific training scheme; evidence is internal ablations and DualTalk metrics, not an external theory or dataset of pure interaction labels.

pith-pipeline@v1.1.0-grok45 · 26529 in / 3329 out tokens · 36601 ms · 2026-07-14T12:41:51.217905+00:00 · methodology

0 comments

read the original abstract

Dyadic conversational motion generation is essential for realistic interactive digital humans. Existing approaches typically model conversational behaviors within unified dyadic generators. However, such holistic formulations tend to couple self-speech-driven motion with partner-responsive social feedback, leaving the interaction-specific component implicit and underutilizing the speech-motion correspondence already learned by pretrained monologic motion models. We propose Learn2Chat, a unified framework that models dyadic motion as interaction modulation over pretrained monologic motion priors. This design separates intrinsic speech-driven motion from social interaction effects and enables more structured interaction modeling. Specifically, we introduce a Monologic-Anchored Motion Factorization scheme that leverages the semantic motion manifold learned from monologic data to disentangle audio-driven motion dynamics from interaction-induced modulation, yielding clean interaction representations from dyadic sequences. On top of this representation space, a Cross-Attentive Interaction Latent Prediction module maps paired speech signals to interaction latents through cross-branch attention and interaction alignment. During inference, the predicted interaction latents modulate canonical monologic motion to generate coherent and synchronized dyadic behaviors in a data-efficient manner. Extensive experiments on the DualTalk benchmark demonstrate that Learn2Chat achieves state-of-the-art performance across both quantitative metrics and perceptual evaluations. Moreover, the framework is model-agnostic and seamlessly integrates with diverse pretrained monologic motion backbones, highlighting the effectiveness of prior reuse and interaction adaptation for scalable conversational motion generation. More visual results are available on the project page.

Figures

Figures reproduced from arXiv: 2607.10313 by Cheng Xu, Haoxin Yang, Shengfeng He, Siyue Chen, Xuemiao Xu, Yihong Lin, Zikai Huang.

**Figure 1.** Figure 1: Comparison of existing audio-driven head motion generation paradigms. Monologic methods (a) ignore interaction dynamics, while existing dyadic approaches (b) either require auxiliary motion inputs or entangle speech and interaction signals through holistic learning. In contrast, Learn2Chat reformulates dyadic generation as interaction modulation over canonical monologic motion, enabling realistic and tempo… view at source ↗

**Figure 2.** Figure 2: Overview of Learn2Chat. Learn2Chat reformulates dyadic head motion generation as interaction modulation of pretrained monologic motion priors. (a) The Monologic-Anchored Motion Factorization Learning explicitly disentangles speechdriven semantics from interaction. (b) The Cross-Attentive Interaction Latent Prediction module infers interaction latent from dual-speaker audio via a dual-stream encoder, enabl… view at source ↗

**Figure 3.** Figure 3: Qualitative comparison of generated dyadic head motion sequences across different methods. Our approach produces more natural and interaction-coherent behaviors that better align with real conversational dynamics. TABLE IV: Ablation study on motion factorization and interaction latent prediction design. The results demonstrate the effectiveness of the proposed monologic-anchored motion factorization and cr… view at source ↗

**Figure 4.** Figure 4: User study on lip synchronization, interaction naturalness, and overall quality. Our method consistently outperforms the compared methods across all evaluation criteria. C. Ablation Study In this section, we conduct in-depth ablations to validate the contribution of our key design choices. Effectiveness of the Interaction-Modulated Framework. We examine the necessity of interaction modulation by removing… view at source ↗

discussion (0)

Reference graph

Works this paper leans on

14 extracted references · 2 linked inside Pith

[1]

Advances in neural information processing systems33, 12449–12460 (2020)

Baevski, A., Zhou, Y ., Mohamed, A., Auli, M.: wav2vec 2.0: A frame- work for self-supervised learning of speech representations. Advances in neural information processing systems33, 12449–12460 (2020)

2020
[2]

IEEE Transactions on Pattern Analysis and Machine Intelligence (2025)

Chen, Z., Shen, Y ., Cao, L., Zhang, S., Ji, R.: Clip-driven transformer for weakly supervised object localization. IEEE Transactions on Pattern Analysis and Machine Intelligence (2025)

2025
[3]

In: European Conference on Computer Vision

Fan, X., Li, J., Lin, Z., Xiao, W., Yang, L.: Unitalker: Scaling up audio-driven 3d facial animation through a unified model. In: European Conference on Computer Vision. pp. 204–221. Springer (2024)

2024
[4]

In: European Conference on Computer Vision

He, Q., Ji, X., Gong, Y ., Lu, Y ., Diao, Z., Huang, L., Yao, Y ., Zhu, S., Ma, Z., Xu, S., et al.: Emotalk3d: High-fidelity free-view synthesis of emotional 3d talking head. In: European Conference on Computer Vision. pp. 55–72. Springer (2024)

2024
[5]

In: Proceedings of the AAAI conference on artificial intelligence

Kim, J., Cho, J., Park, J., Hwang, S., Kim, D.E., Kim, G., Yu, Y .: Deeptalk: Dynamic emotion embedding for probabilistic speech-driven 3d face animation. In: Proceedings of the AAAI conference on artificial intelligence. vol. 39, pp. 4275–4283 (2025)

2025
[6]

ACM Trans

Li, T., Bolkart, T., Black, M.J., Li, H., Romero, J.: Learning a model of facial shape and expression from 4d scans. ACM Trans. Graph.36(6), 194–1 (2017)

2017
[7]

arXiv preprint arXiv:1805.06334 (2018)

Liebel, L., Körner, M.: Auxiliary tasks in multi-task learning. arXiv preprint arXiv:1805.06334 (2018)

Pith/arXiv arXiv 2018
[8]

arXiv preprint arXiv:1711.05101 (2017)

Loshchilov, I., Hutter, F.: Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101 (2017)

Pith/arXiv arXiv 2017
[9]

In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

Ng, E., Joo, H., Hu, L., Li, H., Darrell, T., Kanazawa, A., Ginosar, S.: Learning to listen: Modeling non-deterministic dyadic facial motion. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 20395–20405 (2022)

2022
[10]

In: Proceed- ings of the Computer Vision and Pattern Recognition Conference

Peng, Z., Fan, Y ., Wu, H., Wang, X., Liu, H., He, J., Fan, Z.: Dualtalk: Dual-speaker interaction for 3d talking head conversations. In: Proceed- ings of the Computer Vision and Pattern Recognition Conference. pp. 21055–21064 (2025)

2025
[11]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

Peng, Z., Hu, W., Shi, Y ., Zhu, X., Zhang, X., Zhao, H., He, J., Liu, H., Fan, Z.: Synctalk: The devil is in the synchronization for talking head synthesis. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 666–676 (2024)

2024
[12]

In: 2025 International Conference on 3D Vision (3DV)

Sun, X., Zhang, L., Zhu, H., Zhang, P., Zhang, B., Ji, X., Zhou, K., Gao, D., Bo, L., Cao, X.: Vividtalk: One-shot audio-driven talking head generation based on 3d hybrid prior. In: 2025 International Conference on 3D Vision (3DV). pp. 713–722. IEEE (2025)

2025
[13]

In: European Conference on Computer Vision

Tran, M., Chang, D., Siniukov, M., Soleymani, M.: Dim: Dyadic interac- tion modeling for social behavior generation. In: European Conference on Computer Vision. pp. 484–503. Springer (2024)

2024
[14]

In: Proceedings of the Computer Vision and Pattern Recognition Con- ference

Wang, H., Weng, Y ., Li, Y ., Guo, Z., Du, J., Niu, S., Ma, J., He, S., Wu, X., Hu, Q., et al.: Emotivetalk: Expressive talking head generation through audio information decoupling and emotional video diffusion. In: Proceedings of the Computer Vision and Pattern Recognition Con- ference. pp. 26212–26221 (2025)

2025

[1] [1]

Advances in neural information processing systems33, 12449–12460 (2020)

Baevski, A., Zhou, Y ., Mohamed, A., Auli, M.: wav2vec 2.0: A frame- work for self-supervised learning of speech representations. Advances in neural information processing systems33, 12449–12460 (2020)

2020

[2] [2]

IEEE Transactions on Pattern Analysis and Machine Intelligence (2025)

Chen, Z., Shen, Y ., Cao, L., Zhang, S., Ji, R.: Clip-driven transformer for weakly supervised object localization. IEEE Transactions on Pattern Analysis and Machine Intelligence (2025)

2025

[3] [3]

In: European Conference on Computer Vision

Fan, X., Li, J., Lin, Z., Xiao, W., Yang, L.: Unitalker: Scaling up audio-driven 3d facial animation through a unified model. In: European Conference on Computer Vision. pp. 204–221. Springer (2024)

2024

[4] [4]

In: European Conference on Computer Vision

He, Q., Ji, X., Gong, Y ., Lu, Y ., Diao, Z., Huang, L., Yao, Y ., Zhu, S., Ma, Z., Xu, S., et al.: Emotalk3d: High-fidelity free-view synthesis of emotional 3d talking head. In: European Conference on Computer Vision. pp. 55–72. Springer (2024)

2024

[5] [5]

In: Proceedings of the AAAI conference on artificial intelligence

Kim, J., Cho, J., Park, J., Hwang, S., Kim, D.E., Kim, G., Yu, Y .: Deeptalk: Dynamic emotion embedding for probabilistic speech-driven 3d face animation. In: Proceedings of the AAAI conference on artificial intelligence. vol. 39, pp. 4275–4283 (2025)

2025

[6] [6]

ACM Trans

Li, T., Bolkart, T., Black, M.J., Li, H., Romero, J.: Learning a model of facial shape and expression from 4d scans. ACM Trans. Graph.36(6), 194–1 (2017)

2017

[7] [7]

arXiv preprint arXiv:1805.06334 (2018)

Liebel, L., Körner, M.: Auxiliary tasks in multi-task learning. arXiv preprint arXiv:1805.06334 (2018)

Pith/arXiv arXiv 2018

[8] [8]

arXiv preprint arXiv:1711.05101 (2017)

Loshchilov, I., Hutter, F.: Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101 (2017)

Pith/arXiv arXiv 2017

[9] [9]

In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

Ng, E., Joo, H., Hu, L., Li, H., Darrell, T., Kanazawa, A., Ginosar, S.: Learning to listen: Modeling non-deterministic dyadic facial motion. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 20395–20405 (2022)

2022

[10] [10]

In: Proceed- ings of the Computer Vision and Pattern Recognition Conference

Peng, Z., Fan, Y ., Wu, H., Wang, X., Liu, H., He, J., Fan, Z.: Dualtalk: Dual-speaker interaction for 3d talking head conversations. In: Proceed- ings of the Computer Vision and Pattern Recognition Conference. pp. 21055–21064 (2025)

2025

[11] [11]

In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

Peng, Z., Hu, W., Shi, Y ., Zhu, X., Zhang, X., Zhao, H., He, J., Liu, H., Fan, Z.: Synctalk: The devil is in the synchronization for talking head synthesis. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 666–676 (2024)

2024

[12] [12]

In: 2025 International Conference on 3D Vision (3DV)

Sun, X., Zhang, L., Zhu, H., Zhang, P., Zhang, B., Ji, X., Zhou, K., Gao, D., Bo, L., Cao, X.: Vividtalk: One-shot audio-driven talking head generation based on 3d hybrid prior. In: 2025 International Conference on 3D Vision (3DV). pp. 713–722. IEEE (2025)

2025

[13] [13]

In: European Conference on Computer Vision

Tran, M., Chang, D., Siniukov, M., Soleymani, M.: Dim: Dyadic interac- tion modeling for social behavior generation. In: European Conference on Computer Vision. pp. 484–503. Springer (2024)

2024

[14] [14]

In: Proceedings of the Computer Vision and Pattern Recognition Con- ference

Wang, H., Weng, Y ., Li, Y ., Guo, Z., Du, J., Niu, S., Ma, J., He, S., Wu, X., Hu, Q., et al.: Emotivetalk: Expressive talking head generation through audio information decoupling and emotional video diffusion. In: Proceedings of the Computer Vision and Pattern Recognition Con- ference. pp. 26212–26221 (2025)

2025