A speech-input LLM that feeds Whisper encoder features through a learned projector into Llama 3.2 1B Instruct, so it can respond to spoken input without a separate transcription step. It replies in text, so pair it with a TTS voice.
nvidia/nemotron-3.5-asr-streaming-0.6b
Indexed from Hugging FaceLinked to its home on Hugging Face. FrontierOS indexes it; nothing is hosted or mirrored here.ModelSpeech to text638M paramsEst. Edge GPUEstimated minimum compute for inference: Jetson Orin class, 8–16 GB shared memory. Not a benchmark.OpenMDW 1.1Custom or unstated terms. The upstream license governs use.
About
A cache-aware FastConformer-RNNT streaming recognizer that covers 40 language-locales from one checkpoint via language-ID prompting, outputs punctuated text, and lets you pick chunk sizes from 80 ms up to about 1.1 s.
Why it matters for physical AI
Streaming, low-latency transcription lets a robot or voice device react while the user is still talking instead of waiting for the utterance to end.
Summary written by FrontierOS in our own words. The upstream card and license govern use. Metadata checked 27 Sep 2026. Read the full card on Hugging Face
Files, weights and version history live on Hugging Face. FrontierOS indexes this repo; it does not host or mirror it.
hf download nvidia/nemotron-3.5-asr-streaming-0.6bUses the Hugging Face CLI (pip install -U huggingface_hub). This downloads the whole repo; add --include "<pattern>" to fetch only the files you need.
More speech to text
Browse allA faster Whisper large-v3 derivative whose decoder is cut from 32 layers to 4, giving multilingual transcription in roughly 100 languages at a small accuracy cost (it was not fine-tuned for translation).
The small size of the Moonshine Streaming family, an English-only encoder-decoder recognizer that uses a sliding-window encoder so it can transcribe live audio with low latency on edge hardware.