A speech transmitting device includes a signal processing unit for carrying out plural types of speech signal processes by using a speech signal in a received packet, an alignment state detecting unit for detecting an alignment state of the received packet, and a process selecting unit for selecting a speech signal process which corresponds to the alignment state being detected by the alignment state detecting means, based on a selection reference used to select the speech signal process in correspondence to the alignment state of the received packet, and then outputting a result which is processed by the signal processing means according to the selected speech signal process.