natural language processing; neural nets; spectral analysis; speech recognition; ASR improvement; CER; CNN; FFV; Mandarin speech recognition improvement; Thai speech recognition improvement; acoustic modeling approach; automatic speech recognition improvement; character error rates; convolutional neural network; effectivenesses evaluation; frequency domain; fundamental frequency variation; lexical meaning; relative WER reductions; spectral correlation modeling; spectral variation reduction; speech signals; tonal languages; tone features; tone information; word error rates; Decision support systems; Radio frequency; Rail to rail inputs;
机译:基于音频核模型和神经网络的普通话连续语音识别
机译:基于音频核模型和神经网络的普通话连续语音识别
机译:基于音频核模型和神经网络的普通话连续语音识别
机译:将音调合并到卷积神经网络中,以改善普通话/泰语语音识别
机译:用于与说话人无关的语音识别的卷积神经网络。
机译:利用深卷积神经网络对语音情感识别的特征选择算法的影响
机译:基于卷积神经网络的语音情感识别: 输入特征,信号长度和作用语音的影响研究