pdf-icon

Whisper-small

Introduction

Whisper is a pre-trained automatic speech recognition (ASR) and speech translation model. The model is trained on 680,000 hours of labeled data and demonstrates strong generalization capabilities across multiple datasets and domains without requiring fine-tuning.

Available NPU Models

whisper-small

Supported Platforms: LLM630 Compute Kit, Module LLM, and Module LLM Kit

  • This model supports multilingual speech recognition and translation.

  • Encoding Time: 1606.08 ms

  • Average Decoding Time: 148.92 ms

Installation

apt install llm-model-whisper-small

Supported Platforms: Raspberry Pi5

Installation

sudo apt install llm-model-whisper-small-axcl

Supported Platforms: AI Pyramid

Installation

apt install llm-model-whisper-small-ax650
On This Page