Converged IP networks seek to incorporate voice, data, and video on the same infrastructure. However, the integration of all types of traffic onto a single IP network has several advantages as well as disadvantages. While reducing cost and increasing mobility and functionality, VoIP may lead to reliability concerns, degraded voice quality, incompatibility, and end-user complaints due to changing network characteristics. Voice quality degrades considerably due to low bandwidth, high packet loss rates, high jitter, or if total end-to-end delay is greater than the ITU-T suggested 150ms. In order to ensure these strict requirements are met, the underlying network must deploy various schemes to ensure resource availability. This research proposes an adaptive codec selection mechanism which changes the voice encoding scheme in the middle of an active call based on the network conditions. The mechanism is mainly proposed for H.323 based systems and is intended to cause little to no effect on voice quality. The proposed mechanism involves establishing a three-way handshake process in mid-call to renegotiate station capabilities, making the switch at a determined sequence number in an RTP packet. The proposed mechanism ensures the voice continuity while switching codecs by filling the play out buffers appropriately. The effect of these changes on voice quality is determined using the objective E-model.