Moonshine Streaming Small -- GGUF

GGUF conversions and quantisations of UsefulSensors/moonshine-streaming-small for use with CrispStrobe/CrispASR.

Available variants

File Quant Size Notes
moonshine-streaming-small.gguf F32 535 MB Full precision
moonshine-streaming-small-q4_k.gguf Q4_K 243 MB Quantized

Model details

  • Architecture: Streaming encoder-decoder ASR. Raw-waveform audio frontend (no mel) + sliding-window transformer encoder (10L, 620d) + autoregressive transformer decoder (10L, 512d, SiLU-gated MLP, partial RoPE)
  • Parameters: 123M
  • Languages: English
  • License: MIT
  • Source: UsefulSensors/moonshine-streaming-small
  • Designed for: Low-latency streaming ASR on edge devices

Usage with CrispASR

./build/bin/crispasr --backend moonshine-streaming -m moonshine-streaming-small-q4_k.gguf -f audio.wav

Notes

  • Tokenizer (tokenizer.bin) must be in the same directory as the model file
  • Streaming architecture: sliding-window attention with 80ms lookahead
  • Audio frontend processes raw waveform (no mel spectrogram needed)
Downloads last month
90
GGUF
Model size
0.1B params
Architecture
moonshine_streaming
Hardware compatibility
Log In to add your hardware

We're not able to determine the quantization variants.

Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for cstr/moonshine-streaming-small-GGUF

Quantized
(2)
this model