tf2-speech-recognition-las

Tensorflow 2 Speech Recognition Code (LAS)

Developers

• Younglo Lee ([email protected])

Contents

• Contents
• Speech Recognition
• Features
• Prerequisites
• Examples
• References

Speech Recognition

• Tensorflow 2 implementation for korean speech recognition.
• Dataset can be downloaded in http://www.aihub.or.kr/aidata/105/download

Features

• Listen, Attend and Spell (LAS) speech recognition model
• Location sensitive attention
• Korean syllable tokenization

Prerequisites

• Python 3.x
• Tensorflow 2.2.0
• Librosa 0.6.3
• etc.

Examples

• Tensorboard examples (trained on data with short length (<=500))

- Attention weight examples

References and Resources

• https://github.com/Rayhane-mamah/Tacotron-2
• https://github.com/clovaai/ClovaCall