A speaker cluster model is used to manage recognition problems caused by between-speaker variability.

• 本发明提出的语音辨识方法系使用语者分群模型来处珀因语者特性的差异所产生的辨识问题。