Skip to content
Other

sherpa-onnx-whisper-small - Whisper model export for ONNX and Unreal Engine

2 min readupdated

This model is a version of the well-known Whisper Small architecture, repackaged into the ONNX format (an open format for representing machine learning models) for use in the Sentients plugin audio core for Unreal Engine. It is designed for speech-to-text translation across 99 languages and optimized for fast performance across various runtimes, including CPU, DirectML, and TensorRT. The package includes both original floating-point (fp32) files and optimized half-precision (fp16) encoder versions. This tool is useful for Unreal Engine game and application developers who need reliable speech recognition directly inside the engine.

Key features

  • Export of the popular OpenAI Whisper Small architecture with support for 99 languages
  • Support for various runtimes: CPU, DirectML, TensorRT, and HLSL
  • Availability of half-precision (fp16) files to speed up encoder performance
  • Integration with voice input components and the Silero VAD plugin for silence filtering

What you can do

  • Implement voice control or speech recognition in Unreal Engine projects via the Sentients plugin
  • Use optimized ONNX files for inference on various hardware platforms without PyTorch dependencies
  • Set up local audio transcription using fixed shapes for faster execution in specialized runtimes

Example

The model is used through Sentients plugin components in Unreal Engine, selected by the identifier EAudioModel::WhisperSmall, accepting CoreAudioInput as input and returning CoreTranscriptOutput along with the VAD module for voice activity detection.

Read moreDetails
Format: ONNX
Precision: fp32 and fp16
Base model: openai/whisper-small (exported via csukuangfj/sherpa-onnx-whisper-small)
Features: fixed shapes for runtimes, SHA-256 file pinning in the Sentients directory
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.5 GB · ready files1.1 GB
  • FP8 / INT8recommendedweights 0.3 GB0.8 GB
  • 4-bitweights 0.1 GB0.7 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # sentientsai/sherpa-onnx-whisper-small
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download sentientsai/sherpa-onnx-whisper-small --local-dir ~/models/sherpa-onnx-whisper-small
    # The model is in ~/models. How to run it is in the model card (link above).
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of sherpa-onnx-whisper-small (quants, finetunes, LoRAs) will show up under the bell.

No known relatives of this model yet. All derivatives on Hugging Face

About the model

sentientsai/sherpa-onnx-whisper-small
Author
sentientsai
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments