A comparison of differential schemes for low-rate sinusoidal audio coding

Sinusoidal coding plays an important part in low rate audio coding. Typically, differential techniques are employed to reduce the bit rate for the representation of sinusoidal parameters. We compare several schemes for time-differential (TD) and frequency-differential (FD) representation of the sinusoidal model parameters. We show, through simulation experiments, that bit rates obtained with FD techniques are competitive with those achieved using TD techniques, provided that a variable length segmentation algorithm is first applied to the input signal. This result is important because FD techniques (in contrast to TD techniques) do not rely on the presence of previous segments for correct decoding, and therefore offer robust performance in a lossy packet channel environment.