Automatic Speech Recognition

Introduction

Automatic Speech Recognition (ASR) models are designed to convert spoken language into text. These models are useful in various applications, including transcription services, voice-controlled systems, and accessibility tools.

Available CPU Models

llm-model-sherpa-ncnn-streaming-zipformer-20m-2023-02-17

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

• This model is a streaming ASR model based on the Zipformer architecture, trained on a large dataset to provide accurate speech recognition capabilities.

• This model only supports English.

Install

apt install llm-model-sherpa-ncnn-streaming-zipformer-20m-2023-02-17

• Download llm-model-sherpa-ncnn-streaming-zipformer-20m-2023-02-17

llm-model-sherpa-ncnn-streaming-zipformer-zh-14m-2023-02-23

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

• This model is a streaming ASR model based on the Zipformer architecture, trained on a large dataset to provide accurate speech recognition capabilities.

• This model only supports Chinese.

Install

apt install llm-model-sherpa-ncnn-streaming-zipformer-zh-14m-2023-02-23

• Download llm-model-sherpa-ncnn-streaming-zipformer-zh-14m-2023-02-23