"Best effort" packet-switched networks, like the Internet, cannot offer a reliable transmission of real-time data such as voice and can lead to excessive packet losses under congestion. In real time packetized voice application, missing packets are the major source of voice quality degradation. Therefore, packet loss concealment (PLC) algorithms are needed to guarantee the QoS (Quality of Service) of VoIP (Voice over Internet Protocol). In this thesis, we describe a packet loss concealment scheme utilizing the next and previous good frames that follows and precedes lost packets, i.e., both-side information-based concealment. It is very natural to expect that this scheme should intuitively produce better concealment results than only the preceding frame information-based approach. The proposed concealment algorithm executes a phoneme comparison and both-side frame information-based waveform substitution to generate more suitable signal for the lost interval. In addition, it can be combined with other one-side PLC schemes such as the LP (Linear Prediction) based PLC algorithm, G.711 pitch waveform replication, recommended by ITU-T. Additional voice quality improvement is also obtained through this integration. Our proposed algorithm was evaluated by subjective quality test and compared with other PLC algorithms. According to the results, it provides more improved sound when the lost interval contains transition region or is too long to be concealed by utilizing only the preceding speech signal information.