Language recognition method and device

The invention provides a language recognition method and device. The recognition method comprises the following steps: respectively extracting a first cepstrum coefficient feature and a second cepstrum coefficient feature from a to-be-recognized voice signal; generating an image feature based on the extracted first cepstrum coefficient feature and the second cepstrum coefficient feature, and converting the image feature into a color visual image; and inputting the features of the color visual image into a deep learning model for language classification recognition, and outputting a recognition result. In this way, the recognition effect and precision can be improved. 本发明提供一种语言识别方法及装置，该识别方法包括：从待识别语音信号中分别提取第一倒谱系数特征及第二倒谱系数特征；基于所提取的第一倒谱系数特征及第二倒谱系数特征生成图像特征，并转为彩色视觉图像；将所述彩色视觉图像的特征输入深度学习模型进行语言分类识别，输出识别结果。通过上述方式，可以提高识别效果及精度。