VoxCPM is an innovative tokenizer-free end-to-end text-to-speech (TTS) system that overcomes the limitations of discrete tokenization by modeling speech in a continuous space. It has two core capabilities: context-aware speech generation and realistic zero-shot voice cloning. It can automatically adjust the prosody and style according to the text content and clone the speaker's timbre, accent, and emotion with just a short reference audio.

Audio ProcessingSafetensorsMultiple Languages

Hunyuan OCR is an end-to-end OCR expert VLM model driven by the Hunyuan native multimodal architecture. With a lightweight design of only 1 billion parameters, it has achieved state-of-the-art results in multiple industry benchmark tests. This model excels in handling complex multilingual document parsing and performs well in practical application scenarios such as text localization, open-domain information extraction, video subtitle extraction, and image translation.

MultimodalTransformersMultiple Languages

Emu3.5 is a native multimodal model developed by the Beijing Academy of Artificial Intelligence (BAAI). It can jointly predict the next state across vision and language, enabling coherent world modeling and generation. After end-to-end pre-training and large-scale reinforcement learning post-training, it demonstrates excellent performance in multimodal tasks.

MultimodalTransformers

LightOnOCR-1B-1025 is a compact end-to-end vision-language model specifically designed for optical character recognition and document understanding. It achieves state-of-the-art accuracy at the same weight level, while being faster and less costly than large general-purpose vision-language models.

MultimodalSafetensorsMultiple Languages

OpenCUA is an end-to-end computer usage foundation model series, built on the Qwen2.5-VL instruction model, capable of generating executable operations in a computer environment. It has powerful visual positioning and multi-step task planning capabilities, and performs excellently in computer usage agent benchmark tests such as OSWorld.

MultimodalTransformersEnglish

Ming-UniAudio is an innovative unified speech framework that integrates speech understanding, generation, and editing functions into one. It uses a unified continuous speech tokenizer, which can effectively fuse semantic and acoustic features in an end-to-end model. It is the first system that can achieve general and free-form speech editing only through natural language instructions.

MultimodalSafetensors

MapAnything is an end-to-end trained Transformer model that can take multiple modalities as input and directly regress the decomposed metric 3D geometric structure of the scene. This model supports more than 12 different 3D reconstruction tasks, including multi-image SfM, multi-view stereo vision, and monocular metric depth estimation.

Computer VisionSafetensorsEnglish

Step-Audio 2 is an end-to-end multimodal large language model designed to meet the industry-level audio understanding and voice dialogue needs. It has advanced voice and audio understanding capabilities, intelligent voice dialogue functions, tool invocation, and multimodal retrieval enhanced generation capabilities, and has achieved leading performance in multiple audio understanding and dialogue benchmark tests.

MultimodalTransformers

LFM2-Audio-1.5B is the first end-to-end audio foundation model launched by Liquid AI, designed specifically for low-latency and real-time conversations. With only 1.5 billion parameters, this model can achieve seamless conversational interactions, and its capabilities can rival those of models with much larger parameter scales.

MultimodalSafetensorsEnglish

ARC-Hunyuan-Video-7B is a multimodal model developed by Tencent ARC Lab, specifically designed to understand short video content in the real world. It can process visual, audio, and text signals end-to-end, achieving in-depth structured understanding of videos by integrating multimodal clues.

MultimodalTransformers

Qwen2.5-Omni-3B-GGUF is an end-to-end multimodal model capable of perceiving various modal information such as text, images, audio, and video, and generating text and natural speech responses in a streaming manner.

MultimodalTransformersEnglish

Qwen2.5-Omni is an end-to-end multimodal model capable of perceiving various modalities such as text, images, audio, and video, and generating text and natural speech responses in a streaming manner.

MultimodalTransformersEnglish

Qwen2.5-Omni is an end-to-end multimodal model capable of perceiving multiple modalities including text, images, audio, and video, while generating text and natural speech responses in a streaming manner.

MultimodalTransformersEnglish

Qwen2.5-Omni is an end-to-end multimodal model capable of perceiving various modalities including text, images, audio, and video, while synchronously generating text and natural speech responses in a streaming manner.

MultimodalTransformersEnglish

EMOVA is an end-to-end omni-modal large language model that supports visual, auditory, and speech functions, capable of generating text and speech responses with emotional control.

MultimodalTransformersMultiple Languages

Qwen2.5-Omni is an end-to-end multimodal model capable of perceiving various modalities such as text, images, audio, and video, and generating text and natural speech responses in a streaming manner.

MultimodalTransformersEnglish

EMOVA is an end-to-end omni-modal large language model that supports visual, auditory, and speech capabilities, with emotional speech dialogue abilities.

MultimodalTransformersMultiple Languages

EMOVA is an end-to-end omnipotent modality large language model that supports vision, hearing, and speech functions, enabling multimodal understanding and generation without relying on external models.

MultimodalTransformersMultiple Languages

Advanced model for end-to-end synthetic song detection, capable of identifying AI-generated complete songs (including vocals, music, lyrics, and style)

Audio ProcessingPytorchEnglish

An advanced model for detecting end-to-end AI-generated songs, particularly adept at identifying long-term forged audio features

Audio ProcessingPytorchEnglish