Whispering in Norwegian: Navigating Orthographic and Dialectic Challenges
It addresses orthographic and dialectic challenges in Norwegian ASR, providing a domain-specific solution for improved transcription accuracy.
The paper tackles Norwegian speech recognition by fine-tuning OpenAI's Whisper to create NB-Whisper, improving word error rates from 10.4 to 6.6 on Fleurs and from 6.8 to 2.2 on NST datasets.
This article introduces NB-Whisper, an adaptation of OpenAI's Whisper, specifically fine-tuned for Norwegian language Automatic Speech Recognition (ASR). We highlight its key contributions and summarise the results achieved in converting spoken Norwegian into written forms and translating other languages into Norwegian. We show that we are able to improve the Norwegian Bokmål transcription by OpenAI Whisper Large-v3 from a WER of 10.4 to 6.6 on the Fleurs Dataset and from 6.8 to 2.2 on the NST dataset.