Search papers, labs, and topics across Lattice.
1
0
3
Ditch slow, sequential decoding: NLE achieves 27x speedup over autoregressive ASR by using a non-autoregressive, LLM-based transcript editing approach.
