Wei Liu

Achieve efficient and positionally consistent simultaneous machine translation with LLMs, regardless of the positional encoding method, using a surprisingly simple explicit position allocation strategy.

Reviving degraded opera videos gets a boost: TextOVSR leverages text prompts to guide super-resolution, outperforming state-of-the-art methods by explicitly modeling both degradation and content semantics.

Ditch the noise priors: a new neural beamforming approach uses cross-attention to extract target sources from complex audio mixtures, outperforming prior art that requires explicit noise estimation.