gothic has three nasal consonants, of which one is an allophone of the others, found only in.
哥德语有三个鼻音,其中一个是另外两个的同位异音,只算是它们的。
the experiment results indicate that the training of context-independent phone models and of allophone models are reliable using this method.
实验结果表明,使用给出的方法,可以实现对与发音语境的无关音素模型,以及异音模型的可靠训练。
before training distribution-shared allophone models for continuous speech recognition, it is context-independent phone models that have to be trained.
在语音识别中,为了得到分布共享的异音模型,先要知道与发音语境无关的音素模型。