Saved in:
| Main Authors: | , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2501.04561 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912600448565248 |
|---|---|
| author | Luo, Run Lin, Ting-En Zhang, Haonan Wu, Yuchuan Liu, Xiong Yang, Min Li, Yongbin Chen, Longze Li, Jiaming Zhang, Lei Xia, Xiaobo Alinejad-Rokny, Hamid Huang, Fei |
| author_facet | Luo, Run Lin, Ting-En Zhang, Haonan Wu, Yuchuan Liu, Xiong Yang, Min Li, Yongbin Chen, Longze Li, Jiaming Zhang, Lei Xia, Xiaobo Alinejad-Rokny, Hamid Huang, Fei |
| contents | Recent advancements in omnimodal learning have significantly improved understanding and generation across images, text, and speech, yet these developments remain predominantly confined to proprietary models. The lack of high-quality omnimodal datasets and the challenges of real-time emotional speech synthesis have notably hindered progress in open-source research. To address these limitations, we introduce \name, a two-stage training framework that integrates omnimodal alignment and speech generation to develop a state-of-the-art omnimodal large language model. In the alignment phase, a pre-trained speech model undergoes further training on text-image tasks, enabling (near) zero-shot generalization from vision to speech, outperforming models trained on tri-modal datasets. In the speech generation phase, a lightweight decoder is trained on speech tasks with direct preference optimization, enabling real-time emotional speech synthesis with high fidelity. Experiments show that \name surpasses state-of-the-art models across omnimodal, vision-language, and speech-language benchmarks. It achieves a 4-point absolute improvement on OmniBench over the leading open-source model VITA, despite using 5x fewer training samples and a smaller model size (7B vs. 7x8B). Additionally, \name achieves real-time speech generation with <1s latency at non-autoregressive mode, reducing inference time by 5x compared to autoregressive methods, and improves emotion classification accuracy by 7.7\% |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_04561 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | OpenOmni: Advancing Open-Source Omnimodal Large Language Models with Progressive Multimodal Alignment and Real-Time Self-Aware Emotional Speech Synthesis Luo, Run Lin, Ting-En Zhang, Haonan Wu, Yuchuan Liu, Xiong Yang, Min Li, Yongbin Chen, Longze Li, Jiaming Zhang, Lei Xia, Xiaobo Alinejad-Rokny, Hamid Huang, Fei Computation and Language Computer Vision and Pattern Recognition Recent advancements in omnimodal learning have significantly improved understanding and generation across images, text, and speech, yet these developments remain predominantly confined to proprietary models. The lack of high-quality omnimodal datasets and the challenges of real-time emotional speech synthesis have notably hindered progress in open-source research. To address these limitations, we introduce \name, a two-stage training framework that integrates omnimodal alignment and speech generation to develop a state-of-the-art omnimodal large language model. In the alignment phase, a pre-trained speech model undergoes further training on text-image tasks, enabling (near) zero-shot generalization from vision to speech, outperforming models trained on tri-modal datasets. In the speech generation phase, a lightweight decoder is trained on speech tasks with direct preference optimization, enabling real-time emotional speech synthesis with high fidelity. Experiments show that \name surpasses state-of-the-art models across omnimodal, vision-language, and speech-language benchmarks. It achieves a 4-point absolute improvement on OmniBench over the leading open-source model VITA, despite using 5x fewer training samples and a smaller model size (7B vs. 7x8B). Additionally, \name achieves real-time speech generation with <1s latency at non-autoregressive mode, reducing inference time by 5x compared to autoregressive methods, and improves emotion classification accuracy by 7.7\% |
| title | OpenOmni: Advancing Open-Source Omnimodal Large Language Models with Progressive Multimodal Alignment and Real-Time Self-Aware Emotional Speech Synthesis |
| topic | Computation and Language Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2501.04561 |