Voice data at 9600 bps or 14400 bps (full rate) is first passed through a convolutional encoder, which doubles the data rate for the 9600 bps case or increases it by 1.33 times for the 14400 bps case.

It is then interleaved, a process that has no effect on the rate, but does introduce time delays in the final reconstruction of the signal.  The interleaving processes increases the effectiveness of the convolutional encoder. A long code is XOR’ed with the data, which is a voice privacy function and not needed for channelization.

The closed loop power control data is then punctured into the data stream using the long code to determine the exact location of the power control bits.  CDMA then applies a 64 bit Walsh code which is uniquely assigned to a base to mobile link to form one channel.

This sets a physical limit of 64 channels on the forward link. If the coded voice data is a zero, the Walsh sequence is output; if the data is a one, the logical not of the Walsh code is sent. The Walsh coding yields a data rate increase of 64 times. The data is then split into I and Q channels, and scrambled with short codes.  The final signals are passed through a low pass filter, and eventually sent to an I/Q modulator.