-
如何构建一个端到端的自动语音识别系统:使用PyTorch编写的简单指南,例如speech-recognition-examples
资源介绍
使用CRNN,CTC丢失,DeepSpeech波束搜索和KenLM记分器进行语音识别
安装
cd venv/bin
./pip install -r ../../requirements.txt
./pip install deepspeed==0.3.13
建筑学
SpeechRecognitionModel (
# First convolutional layer
( cnn ): Conv2d ( 1 , 32 , kernel_size = ( 3 , 3 ), stride = ( 2 , 2 ), padding = ( 1 , 1 ))
# 7 blocks of convolutional layers with residual connections
( res_cnn ): Sequential (
( 0 ): ResidualCNN