Fact · Speech recognition
NVIDIA's Nemotron Speech Streaming EN 0.6B model uses a cache-aware FastConformer encoder with 24 layers and an RNN-T decoder, which lets it stream audio as it arrives.
Checked against its source on October 17, 2026
In the source's own words
“architecture with 24 encoder layers and an RNNT (Recurrent Neural Network Transducer) decoder”
Nemotron Speech Streaming EN 0.6B model card, NVIDIA, Hugging Face, n.d., Model architecture.
Open the original- What it measures
- The architecture section of the card, which NVIDIA describes as built for real-time speech-to-text.
- What it does not show
- The model is English only. The card's metadata lists its license as 'other', so read its full terms before commercial use.