Model overview
parakeet-tdt-0.6b is described by NVIDIA as a multilingual automatic speech recognition model. It handles speech across 25 European languages and returns transcripts that include punctuation and capitalisation. The model also produces word-level timestamps alongside the transcribed text.
- Multilingual ASR across 25 European languages
- Punctuation and capitalisation in output
- Word timestamps
Availability
The model appears in the NVIDIA model catalogue on build.nvidia.com, which lists models published by NVIDIA for building generative AI applications. parakeet-tdt-0.6b carries the "Downloadable" label in that catalogue. It is listed alongside other NVIDIA speech and language models, including a translation model and an English real-time speech recognition model.
- Listed in the NVIDIA model catalogue on build.nvidia.com
- Marked as Downloadable
Catalogue context
NVIDIA's model listing page groups more than sixty models alongside a separate set of blueprints. The catalogue spans speech, vision-language, embedding, optical character recognition and video processing models. parakeet-tdt-0.6b sits within this listing as the entry for multilingual speech transcription.
- Part of a catalogue of NVIDIA models and blueprints
- Grouped with other speech, vision and language models
Sources
Last verified 02 Oct 2026. This entry is compiled from the public web pages listed above. Nothing here is stated that those pages do not, and each of them was read on the date shown.