BLANC: Adaptive Temporal Scheduling for Sign Language Production

Youngmin Kim; Seoyeong Hwang; Sujung Hong; Seong Jae Hwang

cs.CVOct 16, 2026

BLANC: Adaptive Temporal Scheduling for Sign Language Production

Authors: Youngmin Kim, Seoyeong Hwang, Sujung Hong, Seong Jae Hwang

Organizations: Yonsei University · Ewha Womans University

Abstract

Sign language production (SLP) aims to synthesize expressive signing motion from written text. Modeling signing motion presents two key temporal challenges. First, fine-grained hand articulations demand higher temporal resolution than relatively slower body movements. Second, signing events unfold over variable durations, calling for adaptive temporal scheduling during generation. To address these challenges, we first introduce a Multi-Rate Vector Quantizer (MRVQ), an articulation-decoupled motion tokenizer that represents the body and both hands as separate discrete streams at distinct temporal resolutions. Building on this representation, we propose BLANC (BLock-wise Adaptive diffusioN with Articulatory Clock), a gloss-free SLP framework that generates variable-duration motion blocks causally while denoising tokens in parallel within each block. At its core, a Latent Articulatory Clock (LAC) learns temporal structure from multi-scale motion-derived partitions to predict block durations and sequence termination during generation. We further introduce inference-time body-conditioned hand refinement to improve coordination across articulators. Extensive experiments on PHOENIX-2014T and CSL-Daily demonstrate that BLANC achieves state-of-the-art performance in sign language production.

Figures & tables

Figure 1: Temporal generation strategies and boundary alignment. (a) Comparison of temporal generation strategies with different generation units and scheduling schemes. (b) Boundary alignment of uniform and Blanc with sign transitions estimated by Cheng et al. (2020) .

Figure 2: Velocity analysis on PHOENIX-2014T and CSL-Daily. Hands exhibit higher clip-level RMS velocity than the body in both global 3D joint space (a) and local angular space (b), motivating separate temporal rates for body and hand motion.

Figure 3: Overview of our method. (a) MRVQ architecture for multi-rate motion tokenization. (b) Training pipeline of Blanc with motion-derived temporal supervision. (c) Inference pipeline with adaptive block generation and body-conditioned hand refinement.

Method	PHOENIX-2014T						CSL-Daily
	DTW-PA-JPE $\downarrow$		DTW-JPE $\downarrow$		B-T $\uparrow$		DTW-PA-JPE $\downarrow$		DTW-JPE $\downarrow$		B-T $\uparrow$
	Body	Hand	Body	Hand	BLEU-4	ROUGE-L	Body	Hand	Body	Hand	BLEU-4	ROUGE-L
T2M † Stoll et al. (2022)	36.54	3.63	35.69	24.12	0.31	5.95	28.83	3.58	30.97	18.63	0.00	10.80
MotionGPT Jiang et al. (2023)	8.22	1.70	8.31	9.07	5.62	16.88	10.05	2.19	9.86	10.65	0.00	9.15
MoMask Guo et al. (2024)	8.20	1.58	7.91	7.69	6.44	19.81	10.58	2.15	10.51	9.61	0.57	15.89
SOKE (no dict.) Zuo et al. (2025)	8.24	1.64	8.69	8.07	5.13	23.86	9.94	1.77	9.63	8.70	2.66	21.13

Table 1: Comparison with gloss-free SLP methods on PHOENIX-2014T and CSL-Daily. B-T denotes back-translation. Bold and underlined indicate the best and second-best results, respectively.

\dagger

marks reproduced results.

Figure 4: Qualitative comparisons with SOKE and SignSparK on PHOENIX-2014T (top) and CSL-Daily (bottom). Each example shows the input text, reference video frames, and generated signing poses.

Method	Ph-T (JPE) $\downarrow$		CSL (JPE) $\downarrow$
Method	Body	Hands	Body	Hands
MotionGPT Jiang et al. (2023)	29.78	5.69	30.71	4.27
MoMask Guo et al. (2024)	23.65	5.27	25.07	4.12
SOKE Zuo et al. (2025)	25.63	4.66	21.56	3.41
Ours (MRVQ)	15.48	4.37	11.80	3.06

Table 2: Reconstruction errors of the tokenizer on PHOENIX-2014T (Ph-T) and CSL-Daily (CSL). We report PA-MPJPE to assess reconstruction performance.

Block Size ( $L^{\prime}$ )	Ph-T			CSL
	DTW-PA-JPE $\downarrow$		B-T $\uparrow$	DTW-PA-JPE $\downarrow$		B-T $\uparrow$
	Body	Hands	BLEU-4	Body	Hands	BLEU-4
$L^{\prime}=1$ (AR)	9.10	1.69	6.37	11.14	2.01	0.68
$L^{\prime}=5$	11.33	2.10	4.37	15.17	2.59	0.98
$L^{\prime}=10$	11.79	2.23	4.10	14.79	2.56	0.98
$L^{\prime}=20$	10.95	2.11	4.76	14.53	2.56	1.18

Table 3: Comparison of fixed block sizes and our adaptive block generation.

L^{\prime}

denotes the fixed block size, with

L^{\prime}=1

corresponding to autoregressive (AR) generation.

PHOENIX-2014T
Temporal Supervision	DTW-PA-JPE $\downarrow$		B-T $\uparrow$	Boundary $\uparrow$
Temporal Supervision	Body	Hands	BLEU-4	Precision	F1
Uniform	7.93	1.51	8.98	55.49	37.89
Shuffled	7.63	1.50	8.54	54.57	37.26
Single-Scale	7.58	1.48	8.44	60.14	41.07
Multi-Scale (Ours)	7.58	1.48	9.85	62.06	42.37

Table 4: Ablation of temporal supervision strategies. Boundary precision. and F1 evaluate alignment with CSLR-derived boundaries.

LAC Component Ablation
Setting	Ph-T			CSL
	DTW-PA-JPE $\downarrow$		B-T $\uparrow$	DTW-PA-JPE $\downarrow$		B-T $\uparrow$
	Body	Hands	BLEU-4	Body	Hands	BLEU-4
w/o Motion Feedback	7.81	1.54	9.64	9.87	1.91	0.23
w/o Alignment Loss	7.61	1.49	9.54	9.79	1.72	1.54
Refinement Method

Table 5: Ablation of LAC components and refinement strategies. For refinement,

\blacktriangle

\blacktriangledown

, and denote improvement, degradation, and no change relative to w/o refinement, respectively.

Appendix figures & tables8 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 5: Qualitative results between ground truth video frames and reconstructed meshes obtained from the proposed our fitting pipeline on Phoenix-2014T (yellow) and CSL-Daily (red).

Method	Body $\downarrow$	Left Hand $\downarrow$	Right Hand $\downarrow$
FrankMoCap Rong et al. (2020)	78.07	20.47	19.62
PIXIE Feng et al. (2021)	60.11	25.02	22.42
PyMAF-X Zhang et al. (2023)	68.61	21.46	19.19
SMPLify-X Pavlakos et al. (2019)	56.07	22.23	18.83
SGNify Forte et al. (2023)	55.63	19.22	17.50
OSX Lin et al. (2023)	47.32	18.34	18.12

Appendix

Table 6: Reconstruction errors on SGNify mocap dataset Forte et al. (2023) . We report mean per vertex errors in mm.

Loss coefficient	Ph-T	CSL
$\lambda_{d}$	0.20	0.25
$\lambda_{\mathrm{len}}$	0.25	0.75
$\lambda_{e}$	0.10	0.10
$\lambda_{a}$	0.10	0.10
$\lambda_{cf}$	0.05	0.05

Appendix

Table 7: Maximum training-loss weights for Blanc on PHOENIX-2014T (Ph-T) and CSL-Daily (CSL). The aggregate motion objective has unit weight in Equation 10 .

Hyperparameter	Value
Hidden / code dimension	512 / 512
Transformer layers / heads	4 / 8
Temporal compression (body / hands)	4 / 2
Codebook size (body / each hand)	96 / 192
Training window (frames)	64
Optimizer	AdamW

Appendix

Table 8: MRVQ hyperparameters, shared across PHOENIX-2014T and CSL-Daily. Transformer settings apply to both the encoder and decoder. L and R denote the left and right hands.

Hyperparameter	Ph-T	CSL
Generator dimension	768
Generator layers / heads	12 / 12
LAC hidden / event dimension	512 / 256
Dropout	0.1
Block duration (body-rate units)	2–11	2–18
MS-Teacher radii	$\{1,2,4\}$

Appendix

Table 9: Blanc hyperparameters on PHOENIX-2014T (Ph-T) and CSL-Daily (CSL). One body-rate unit corresponds to four motion frames. Refinement is applied only at inference.

Figure 6: Qualitative comparison of motion reconstruction on PHOENIX-2014T (blue) and CSL-Daily (yellow). Red circles highlight differences in hand positioning and fine-grained articulation across methods.

Feature Space
Method	Ph-T $\uparrow$		CSL $\uparrow$
Method	BLEU-4	ROUGE-L	BLEU-4	ROUGE-L
GT	13.36	31.39	7.35	31.09
Token Space
MotionGPT	9.95	25.67	4.97	25.81
MoMask	10.07	26.37	5.04	26.10

Appendix

Table 10: Back-translation performance of motion tokenizers. GT denotes the feature-level reference without tokenization.

Figure 7: Effect of the hand refinement ratio on BLEU-4 (left; higher is better) and overall DTW-PA-JPE (right; lower is better). Blue curves use the left axes for PHOENIX-2014T, and orange curves use the right axes for CSL-Daily. Boxes mark our selected ratios: 100% for PHOENIX-2014T and 15% for CSL-Daily.

Sign Language Generation

Sign language generation creates signed motion or video from text, glosses, or other inputs while preserving linguistic structure and natural movement.

See all 34 papers

SignSparK: Efficient Multilingual Sign Language Production via Sparse Keyframe Learning
Mar 11, 2026Jianhe Low, Alexandre Symeonidis-Herzig, Maksym Ivashechkin +2Human Motion Generation Sign Language Generation
CVSSP, University of Surrey, United Kingdom
Conditional Collapse in Sign Language Production: A Diagnostic and a Scaling Argument
Jun 1, 2026Rui Hong, Jana KošeckáSign Language Translation Human Motion Generation
George Mason University
SignRR: Retrieve and Refine Real Motion for Sign Language Production
Aug 28, 2026Fidel Omar Tito Cruz, Angie Sanchez Marquina, Summy Farfan +1Human Motion Generation Motion Representation Learning
University of Central Florida, Florida, USA · Universidad Nacional Mayor de San Marcos, Lima, Peru · Universidad Catolica San Pablo, Arequipa, Peru +1

Mar 11, 2026J/K move · Enter open · S save

SignSparK: Efficient Multilingual Sign Language Production via Sparse Keyframe Learning

Jianhe Low, Alexandre Symeonidis-Herzig, Maksym Ivashechkin, Ozge Mercanoglu Sincan+1

CVSSP, University of Surrey, United Kingdom

Human Motion Generation Sign Language Generation

Open paper AskPDF arXiv

Sign Language Production (SLP) faces a fundamental trade-off: direct text-to-pose models suffer from regression-to-the-mean effects, while dictionary-retrieval methods produce disjointed transitions. To resolve this, we propose a novel training paradigm that leverages sparse keyframes to capture the underlying kinematic distribution of human signing. By generating dense motion from discrete anchors, our approach mitigates regression-to-the-mean while ensuring fluid articulation. To achieve this at scale, we introduce FAST, an ultra-efficient sign segmentation model that automatically mines precise temporal boundaries. We then present SignSparK, a Conditional Flow Matching (CFM) framework that utilizes these temporal anchors to synthesize 3D signing sequences. This keyframe-driven formulation also unlocks Keyframe-to-Pose (KF2P) generation, making precise spatiotemporal editing of signing sequences possible. Furthermore, SignSparK scales across four distinct sign languages, constituting the largest multilingual SLP framework to date, and integrates 3D Gaussian Splatting for photorealistic rendering. Extensive evaluations demonstrate that SignSparK achieves state-of-the-art across diverse SLP tasks and multilingual benchmarks. Our code is available at https://github.com/JianHe0628/SignSparK.

Abstract

Sign Language Generation

Explore similar work

SignSparK: Efficient Multilingual Sign Language Production via Sparse Keyframe Learning

Conditional Collapse in Sign Language Production: A Diagnostic and a Scaling Argument

SignRR: Retrieve and Refine Real Motion for Sign Language Production