System and method for handling underrun of compressed speech frames due

Abstract

A system and method for compensating for slight variances between a first rate at which a far end modem transmits a series of data frames to a near end modem and a second rate at which the near end modem requires reception of the series of data frames. The modems comprise a data pump which receives data frames from the other modem at the first rate on a telephone line. The data frames are received by a data pump which provides the frames a character at a time to a controller coupled to the data pump. The controller places the characters into a pair of frame buffers in a memory coupled to the controller. The controller receives data frames from a vocoder at the second rate to be transmitted to the other modem. The controller attempts to provide data frames to the vocoder at the second rate in order to avoid distortion in the data frame flow, in particular, distortion in the compressed speech. The method comprises determining according to the second rate if a next data frame in the series of data frames has been received and providing the next data frame to a decoder if the next data frame has been received, but providing a filler frame to the decoder if the next data frame has not been received. The modem repeatedly performs these steps for each of the data frames in the series of frames. Preferably the data frames are compressed speech frames and the decoder is a vocoder for decoding the compressed speech frames. Preferably, the filler frame comprises a quiet frame which has been produced by recording essentially silence, i.e., only white noise. Preferably, the modems comprise DSVD modems. The modem also determines if the next data frame is bad if the data frame has been received and provides a filler frame to the decoder if the next data frame is bad.