Sequence uses our custom conversion of the latest Parakeet model from @nvidia, in our testing we managed to the WER down to 2.54 on the LibriSpeech benchmark - done by utilizing extra layers of sound detection on top of what the model already does.

Back to Feed
Sequence uses our custom conversion of the latest Parakeet model from @nvidia, in our testing we managed to the WER down to 2.54 on the LibriSpeech benchmark - done by utilizing extra layers of sound detection on top of what the model already does.

Sequence uses our custom conversion of the latest Parakeet model from @nvidia, in our testing we managed to the WER down to 2.54 on the LibriSpeech benchmark - done by utilizing extra layers of sound detection on top of what the model already does.