Search papers, labs, and topics across Lattice.
1
0
3
Fine-tuning LLMs directly on speech can lead to powerful ASR systems, but requires careful tuning of label units, architectures, and training strategies to avoid hallucinations.
