Page 218 of 430 Nortel IP Softphone 2050553-3001-368 Standard 7.00 August 2005Jitter bufferThe jitter buffer of the IP Softphone 2050 is configurable. Set the jitter bufferin one of the following ways:• using the default value sent from the TPS (the value configured in OTM— It is recommended that the default value be used.)• using the Configuration UtilityThe jitter buffer has a desired size and a maximum allowable size. If the jitterbuffer exceeds its maximum allowable size, sufficient frames are discardedto reduce the contents of the jitter buffer to the desired setting. If the jitterbuffer underruns, frames are held in the jitter buffer until it fills to the desiredlevel. Both underrun and overrun result in a discontinuity in the audio.For codecs that support silence suppression, the jitter buffer is resynchronizedat the beginning of each talk spurt.QoSA combination of codec selection, jitter buffer and packet time, and the useof the network’s DiffServ code point all contribute to the end-to-end QoS.However, the IP Softphone 2050 is an application within the context of thePC OS, so the OS has an effect on the end-to-end QoS for the IPSoftphone 2050. The DSP functionality (such as codec packetization)implemented in DSP hardware on the IP Softphone 2050 and Voice GatewayMedia Card runs as part of the application code on the PC’s CPU. If the CPUis busy with other tasks, voice quality can be negatively affected.The number of buffers used to buffer audio data between the application andPC audio hardware device driver is adjustable from the Configuration Tool(see “Select Sound Devices tab” on page 228). Using fewer buffers reducesthe audio path delay but increases the chances of dropouts and choppy speech,depending on the speed and utilization of the PC’s CPU.DiffSERV (DSCP)The IP Softphone 2050 uses DSCP settings assigned by the TPS. The IPSoftphone 2050 supports DSCP on Window 98, Windows 98 SE, Windows2000 Professional, and Windows XP.