To actualize an editing method for voice-encoded data which reduces deterioration in sound quality due to quantization noise and eliminates the loss of attached information when the bit rate of an encoded acoustic signal is converted and header information, etc., is edited.

When a compressed audio signal bit stream is re-encoded or attached information is edited, data of a part protected by the error detection code of the bit stream is detected and stored in a memory device 40, and the re-encoded bit stream or the bit stream obtained through the editing of an edited data generating means 22 is multiplexed with the error detection code stored in the memory device 40 to generate an encoded bit stream having been edited. Consequently, the editing method for voice-encoded data is actualized and thereby, the editing method is hardly affected by deterioration in sound quality due to quantization noise, eliminates the loss of the attached information, and can obtain the encoded audio signal of high quality.