3

BEAVER is a training-free hierarchical prompt compression framework for long-context LLM inference with strong benchmark performance and 26.4x lower latency on 128k contexts.

We probe widely used speech tokenizers and show they encode phonetic structure much more strongly than lexical-semantic content.

We propose Hive, a semantically consistent 2.4k-hour dataset for data-efficient universal sound separation with strong zero-shot generalization.

Mio is an end-to-end multimodal interactive digital-human framework that combines reasoning and real-time embodiment with state-of-the-art performance.

We introduce a novel token-based text-to-speech (TTS) model with 0.8B parameters, trained on a mix of real and synthetic data totaling 650k hours, to address issues like pronunciation accuracy and style consistency. This model integrates a latent variable sequence with enhanced acoustic information into the TTS system, reducing errors and style changes. Our training includes data augmentation for improved timbre consistency, and we use a few-shot voice conversion model to generate diverse voices. This approach enables learning of one-to-many mappings in speech, ensuring both diversity and timbre consistency. Our model outperforms VALL-E in pronunciation, style maintenance, and timbre continuity.