Differential Vector Quantization (DVQ) is a variable-length lossy compression technique for digital video. An Artificial Neural Network (ANN) is used to develop entropy-biased codebooks which yield substantial data compression without entropy coding and are very robust with respect to transmission channel errors. Huffman coding is a variable-length lossless compression technique where data with a high probability of occurrence is represented with short codewords, while data with lower probability of occurrence is assigned longer codewords. We discuss how these codebooks can be used to realize variable bit-rate coders for the DVQ case and also we address Huffman coding in its extreme effect when data is highly predictable and differential coding can be applied. Two methods are presented for variable bit-rate coding using two different approaches. In the first method, we use DVQ algorithm and both the encoder and the decoder have multiple codebooks of different sizes. In the second, we address the issues of real-time transmission using Huffman coding. The algorithm is based on differential pulse code modulation (DPCM), but additionally utilizes a non-uniform and multi-level Huffman coder to reduce the data rate substantially below that achievable with conventional DPCM. We compare the performance of these two approaches under conditions of error-free and error-prone channels. Our results show that one coding technique is resistant to channel errors than the other, and yields no visible degradation in picture quality at moderate compression rate.