Histogram-Based Quantization for Robust and/or Distributed Speech Recognition