Nova-3 Improved Models for Flemish, German (Switzerland), Lithuanian, and Portuguese
By Steven Van ·
Nova-3 transcription improves for German (Switzerland), Portuguese, Lithuanian and Flemish, with no API changes required.
Deepgram has updated its Nova-3 monolingual speech-to-text models for four languages. German (Switzerland), Portuguese, and Lithuanian get improved batch and streaming models, while Flemish gets an improved streaming model. The updates cover transcription quality and apply automatically: no changes are needed to existing API requests. Self-hosted customers need to request the updated models from their Deepgram account representative to download them.
Full details on the affected models are on the Models & Languages Overview page.
Deepgram
Power enterprise voice solutions with Deepgram’s Speech-to-Text, Text-to-Speech, and Voice Agent APIs. Real-time, accurate, and built for scale.
View Deepgram →