The DVI Adaptive Differential Pulse Code Modulation (ADPCM) algorithm was first described in an IMA recommendation on audio formats and conversion practices [1]. ADPCM is a transformation that encodes 16-bit audio as 4 bits (a 4:1 compression ratio). In order to achieve this level of compression, the algorithm maintains an adaptive value predictor, which uses the distance between previous samples to store the most likely value of the next sample. The difference between samples is quantized down to a new sample using an adaptive step-size. The algorithm in [1] suggests using a table to adapt this step-size to the analyzed data. ADPCM has become widely used and adapted, and a variant of the algorithm performs voice encoding on cellular phones (allowing minimal data to be sent across the wireless network and increasing throughput).