GSM-HR speech codec is currently deemed by us to have no practical utility:

* GSM networks that need to use half-rate channels for capacity reasons will
  preferentially use AMR-HR (superior speech quality to HRv1), and may even
  have to reject GSM MS that don't support it;

* GSM networks that aren't so constrained in capacity will use one of AMR-FR,
  EFR or FRv1, all of which provide better speech quality than HRv1.

Because of this lack of practical utility, some important limits have been
placed on the scope of work to support this codec in the present Themyscira
Wireless GSM codec libraries and utilities suite:

* Between speech encoder & decoder functions on the one hand and the TFO
  transform function (TS 28.062 section C.3.2.1.1) on the other hand, the
  latter received more development attention.  For completeness, TFO transform
  implementations need to be provided for all 3 codecs for which this transform
  is defined, and the implementation of TFO transform for GSM-HR serves as a
  stepping stone toward the much more important one for EFR.

* Librified forms of speech encoder and decoder engines from GSM 06.06 GSM-HR
  reference code have been included in libgsmhr1 for the sake of completeness,
  as well as command line utilities built on top of these library engines.
  However, unlike the situation with EFR and AMR codecs, no effort is being
  expended to make these operations fast or computationally efficient - the
  slow and grossly inefficient code from ETSI remains mostly unchanged except
  for elimination of global variables and introduction of state structures.