SALM: Speech-augmented Language Model with In-context Learning for Speech Recognition and Translation

Open paper
Benchmark Model Rank Results
speech-recognition-on-librispeech-test-cleanNVIDIA Nemotron 3.5 Transcribe 30B A4B#7Word Error Rate (WER): 1.09
speech-recognition-on-spgispeechNVIDIA Nemotron 3.5 Transcribe 30B A4B#8Word Error Rate (WER): 1.91
speech-recognition-on-librispeech-test-otherNVIDIA Nemotron 3.5 Transcribe 30B A4B#15Word Error Rate (WER): 2.6

Rank counts only results with a code link.