Information-theoretic compressibility of speech data

Two standard reversible coding algorithms, Ziv-Lempel and a dynamic Huffman algorithm, are applied to various types of speech data. The data tested were PCM, DPCM, and prediction residuals from LPC. Neither algorithm shows much promise on small amounts of data, but both performed well on large amounts. Typically the Ziv-Lempel required about 12 seconds of data (with 8000 samples per second) to reach a stable compression rate. The dynamic Huffman coding took much less time to "warm up", often needing something like 64 milliseconds. Approximately 66 seconds of PCM with 12 bits per samples was compressed 6.4% by the Ziv-Lempel coding and 20.7% by the dynamic Huffman coding. The same numbers for DPCM with 13 bits per sample are 17.7% and 35.6% respectively. The prediction residuals had compression rates very close to those of DPCM, regardless of whether 1, 2, 5, or 10 prediction coefficients were used.