83
Voted
ICASSP
2011
IEEE
14 years 25 days ago
2011
IEEE
We describe a new approach for rescoring speech lattices — with long-span language models or wide-context acoustic models — that does not entail computationally intensive latt...