Search papers, labs, and topics across Lattice.
Affiliation:
1
0
2
Cached LLM probability retrieval can improve ASR performance without the heavy lifting of retraining, outperforming traditional methods in a majority of tested scenarios.