Quant (x, Xq, QType)

function Index = Quant (x, Xq, QType)
% Quantize a vector of samples
%
% x - vector of input samples
% Xq - Quantizer decision levels (Nreg-1 levels, where Nreg is the number
%    of quantizer intervals).
% QType - Type of quantizer (optional, default 1). These quantizer types
%    differ at the end points of the intervals. For Type 1, if an input
%    value is equal to the upper end of an interval, it is considered to
%    lie in the interval. For Type 2, if an input value is equal to the
%    lower end of an interval, it is consider to lie in the interval.
%
% Index - Output index (0 <= Index < Nreg)
%
%  This function returns the index of the quantizer region corresponding to
%  a given input value. The quantizer is specified by an array of quantizer
%  decision levels. A binary search is used to determine the quantizer
%  region.
%
%  The index value takes on values from 0 to Nreg-1, where Nreg is the
%  number of quantizer regions. If number of regions is equal to one (Xq is
%  empty), the index is set to zero. Otherwise, the index is determined as
%  shown in the following table. The number of decision levels is one less
%  than the number of regions.
%    Index          Interval QType 1              Interval QType 2
%     0             -Inf < x <= Xq(1)            -Inf <= x < Xq(1)
%     1            Xq(1) < x <= Xq(2)           Xq(1) <= x < Xq(2)
%     k            Xq(k) < x <= Xq(k+1)         Xq(k) <= x < Xq(k+1)
%   Nlev-1    Xq(Nlev-1) < x <= Inf        Xq(Nlev-1) <= x < Inf

% For Nlev = 256, this routine is 15 to 20 times faster than quantiz
% (MATLAB Communications Toolbox).

Nreg = length(Xq) + 1;
Index = zeros(size(x));

if (nargin <= 2)
  QType = 1;
end

if (QType == 1)
  [Temp, Index] = histc(-x, -Xq(end:-1:1));
  IU = (x < Xq(1));
  Index(IU) = Nreg-1;
  Index = Nreg - 1 - Index;

else
  [Temp, Index] = histc(x, Xq);
  IU = (x > Xq(end));
  Index(IU) = Nreg-1;
  
end

return