20-1 Introduction

ASRA (Automatic Speech Recognition & Assessment) is a C++ library developed at MIR lab, Taiwan. The goal of this library is to provide preliminary functionalities for

• Speech recognition: Mainly voice command recognition.
• Speech assessment: Also known as pronunciation scoring.

The current version of ASRA supports the above functionalities for the following languages:

• English
• Chinese (Mandarin)
• Japanese
• Taiwanese

However, the support of Japanese and Taiwanese is not as comprehensive as the other two languages.

Moreover, the current version of ASRA supports two platforms, including MS Windows and Unix/Linux.

Note that ASRA only provides APIs for speech recognition and assessment. For training acoustic models, you should refer to HTK tool.

---

Audio Signal Processing and Recognition (音訊處理與辨識)