                 Changes in version 0.2.0 (2026-08-24)                  

  - Removed the subdirectories wave and core from the REAPER build, as
    they're only used for file IO.
  - Made REAPER patches to ensure that all diagnostic messages are
    CRAN-compliant, and added a GetDiagnostics() procedure to capture
    various internally used measures with reaper_wrap() instead of
    writing them to disk.
  - Added several output options to reaper() and reaper_bulk(): gci_cand
    for information about glottal closure instants candidates, probs for
    various pseudo-probabilities relating to voicing, and resids for
    residual waveforms for debugging.
  - Removed the delete parameters from read_epochs_out() and
    read_pitch_out().
  - Refactored reaper_bulk() to allow for new outputs smoothly. Also
    added parameter force_list_output to reaper() as part of this.
  - Added hirst2pass_f0min and hirst2pass_f0max arguments to
    reaper_bulk() for user controlling the pitch floor and ceiling
    values for first pass.
  - Added fileExtension parameter to reaper2emuDB() for selecting the
    extension of SSFF files.
  - Various other minor fixes.

                        Changes in version 0.1.1                        

  - Fixed time domain bug when estimating epochs with start times other
    than 0.
  - Made minimal REAPER patches to fix CRAN check warnings, as well as a
    fix to the StdoutSilencer in the C++ wrapper.
  - Suppress diagnostic messages from REAPER by default.

                        Changes in version 0.1.0                        

  - Previous versions of the library attempted to install the reaper
    command line tool with CMake, and functions would call REAPER from
    the command line and read in output files. This version wraps the
    REAPER C++ source code directly with Rcpp, which should be much less
    fragile. The behavior is otherwise very similar, with a few
    exceptions and breaking changes, including some improved checks.
    There's also some added documentation.
  - reaper() functions largely the same as before, with the addition of
    the arguments start and end for controlling which parts of a sound
    file to analyze, and the channel argument for separating out a
    channel to analyze. This isn't possible with the command line tool.
    The exePath argument has been removed, since the function no longer
    calls an executable. It is now possible to pass files that
    aren't 16-bit (they're converted to 16-bit under the hood).
  - reaper2ssff() has been split into two functions: reaper2ssff(),
    which now only converts single file outputs to the SSFF format and
    stores it in R memory, and reaper2emuDB() for bulk importing to an
    EMU database. This allows users to retain the SSFF objects if they
    e.g. need it for plotting with praatpicture. A bug was spotted and
    fixed in the SSFF conversion process. With reaper2emuDB(), SSFF
    files are now temporarily stored in a temporary directory instead of
    the working directory.
  - The low level reaper_wrap() function has been added which just calls
    a C++ wrapper function and returns a list that isn't in the most
    useful format. A possible advantage of this function is that it
    operates directly on a numeric vector of signal values and doesn't
    require a sound file.
  - The reaper_install() function has been removed.
