Huffman coding technique, which is often used in image processing problems, operates on the symbols that are constructed based on zero numbers of the DCT-transformed and quantized images. In this algorithm, probabilities of the symbols are calculated and codewords are produced for every single symbol. Highest probable symbols are represented in less number of bits. Hence, total number of bits transmitted is minimized. Similarly, it is possible to compress a speech signal that is primarily digitized. In this study, bandwidth change is investigated for a waveform-coding quantized speech signal compressed with Huffman coding technique.