# Multimodal Diffusion Transformer with Memory Bank for Scalable Long-Duration Talking Video Generation

arXiv: https://arxiv.org/abs/2411.16748 · HTML: https://mlsanity.com/paper/2411.16748/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Audio, speech and music, one of 14 directions in Diffusion models.
- Builds on 44 of its 58 references that are in the map.
- First posted on arXiv 24 Nov 2024.

## Placed in

- Diffusion models / [Audio, speech and music](https://mlsanity.com/diffusion/audio-speech-and-music/)

## Builds on (most cited first, 30 of 44)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [High-Resolution Image Synthesis with Latent Diffusion Models](https://mlsanity.com/paper/2112.10752/) (2021)
- [Diffusion Models Beat GANs on Image Synthesis](https://mlsanity.com/paper/2105.05233/) (2021)
- [Scalable Diffusion Models with Transformers](https://mlsanity.com/paper/2212.09748/) (2022)
- [A Style-Based Generator Architecture for Generative Adversarial Networks](https://mlsanity.com/paper/1812.04948/) (2018)
- [Adding Conditional Control to Text-to-Image Diffusion Models](https://mlsanity.com/paper/2302.05543/) (2023)
- [Video Diffusion Models](https://mlsanity.com/paper/2204.03458/) (2022)
- [Transformers are RNNs: Fast Autoregressive Transformers with Linear Attention](https://mlsanity.com/paper/2006.16236/) (2020)
- [AnimateDiff: Animate Your Personalized Text-to-Image Diffusion Models without Specific Tuning](https://mlsanity.com/paper/2307.04725/) (2023)
- [T2I-Adapter: Learning Adapters to Dig out More Controllable Ability for Text-to-Image Diffusion Models](https://mlsanity.com/paper/2302.08453/) (2023)
- [Animate Anyone: Consistent and Controllable Image-to-Video Synthesis for Character Animation](https://mlsanity.com/paper/2311.17117/) (2023)
- [A Lip Sync Expert Is All You Need for Speech to Lip Generation In The Wild](https://mlsanity.com/paper/2008.10010/) (2020)
- [Latte: Latent Diffusion Transformer for Video Generation](https://mlsanity.com/paper/2401.03048/) (2024)
- [VideoCrafter1: Open Diffusion Models for High-Quality Video Generation](https://mlsanity.com/paper/2310.19512/) (2023)
- [Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set](https://mlsanity.com/paper/1903.08527/) (2019)
- [VideoComposer: Compositional Video Synthesis with Motion Controllability](https://mlsanity.com/paper/2306.02018/) (2023)
- [SadTalker: Learning Realistic 3D Motion Coefficients for Stylized Audio-Driven Single Image Talking Face Animation](https://mlsanity.com/paper/2211.12194/) (2022)
- [SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformers](https://mlsanity.com/paper/2410.10629/) (2024)
- [MakeItTalk: Speaker-Aware Talking-Head Animation](https://mlsanity.com/paper/2004.12992/) (2020)
- [Pose-Controllable Talking Face Generation by Implicitly Modularized Audio-Visual Representation](https://mlsanity.com/paper/2104.11116/) (2021)
- [Hierarchical Cross-Modal Talking Face Generationwith Dynamic Pixel-Wise Loss](https://mlsanity.com/paper/1905.03820/) (2019)
- [StreamingT2V: Consistent, Dynamic, and Extendable Long Video Generation from Text](https://mlsanity.com/paper/2403.14773/) (2024)
- [CelebV-HQ: A Large-Scale Video Facial Attributes Dataset](https://mlsanity.com/paper/2207.12393/) (2022)
- [EMO: Emote Portrait Alive -- Generating Expressive Portrait Videos with Audio2Video Diffusion Model under Weak Conditions](https://mlsanity.com/paper/2402.17485/) (2024)
- [Hallo: Hierarchical Audio-Driven Visual Synthesis for Portrait Image Animation](https://mlsanity.com/paper/2406.08801/) (2024)
- [EchoMimic: Lifelike Audio-Driven Portrait Animations through Editable Landmark Conditions](https://mlsanity.com/paper/2407.08136/) (2024)
- [AniPortrait: Audio-Driven Synthesis of Photorealistic Portrait Animation](https://mlsanity.com/paper/2403.17694/) (2024)
- [VASA-1: Lifelike Audio-Driven Talking Faces Generated in Real Time](https://mlsanity.com/paper/2404.10667/) (2024)
- [FreeNoise: Tuning-Free Longer Video Diffusion via Noise Rescheduling](https://mlsanity.com/paper/2310.15169/) (2023)

## Built on by (most cited first, 2 of 2)

- [MuSS: A Large-Scale Dataset and Cinematic Narrative Benchmark for Multi-Shot Subject-to-Video Generation](https://mlsanity.com/paper/2604.23789/) (2026)
- [Efficient Video Diffusion Models: Advancements and Challenges](https://mlsanity.com/paper/2604.15911/) (2026)

## Abstract (arXiv)

Long-duration talking video synthesis faces enduring challenges in achieving high video quality, portrait consistency, temporal coherence, and computational efficiency. As video length increases, issues such as visual degradation, portrait drift, temporal artifacts, and error accumulation become increasingly problematic, severely affecting the realism and reliability of the results. To address these challenges, we present LetsTalk, a diffusion transformer framework equipped with multimodal guidance and a novel memory bank mechanism, explicitly maintaining contextual continuity and enabling robust, high-quality, and efficient generation of long-duration talking videos. In particular, LetsTalk introduces a noise-regularized memory bank to alleviate error accumulation and sampling artifacts during extended video generation. To further improve efficiency and spatiotemporal modeling, LetsTalk employs a deep compression autoencoder and a spatiotemporal-aware transformer with linear attention for effective multimodal fusion. We systematically analyze three fusion schemes and show that combining deep (Symbiotic Fusion) for portrait features and shallow (Direct Fusion) for audio achieves superior visual realism and precise speech-driven motion, while preserving diversity of movements. Extensive experiments demonstrate that LetsTalk establishes new state-of-the-art in generation quality, producing temporally coherent and realistic talking videos with enhanced diversity and liveliness, and maintains remarkable efficiency with 8x fewer parameters than previous approaches.