audiowmark 0.6.3

Overview of Changes in audiowmark-0.6.3:

• update HLS support to build with ffmpeg-7 API (#68)
• preserve wav subformat (#64)
  • write float/double wav output for wav float/double input
  • write signed 16, 24 and 32 bit wav output for 16, 24, 32 bit wav input
  • added test that wav subformat format is preserved (make check)
• improved RawConverter:
  • support float/double input/output (very fast on little endian systems)
  • fix bug for unsigned integer raw input/output (16, 24 and 32 bit)
  • better tests (make check)
  • faster 16/32 bit conversion on little endian systems
• add format wav-pipe for very long input/output streams on stdin/stdout
• performance optimizations for SFOutputStream
• add documentation for Windows/Cygwin builds (#45)

audiowmark 0.6.2

Overview of Changes in audiowmark-0.6.2:

• improved robustness of the watermark detection a lot for many cases
• improved handling if more than one key needs to be used for detection
  • support using --key multiple times for audiowmark get
  • support naming watermark keys (key name defaults to filename)
  • extend JSON and regular output to report key name for matches
• merge architecture documentation for developers written by Tim Janik (#49)
• support RF64 output (--output-format rf64) for huge wav files (#30, #2)
• use hann window function to improve robustness/quality for some files (#48)
• fix building against zita-resampler installed in non-standard location (#39)
• replace sprintf with string_printf: sprintf is deprecated on macOS (#33)
• fix build errors related to PRNG on new clang compilers (#29)
• fix problems in videowmark due to command line option ordering (#23)

audiowmark 0.6.1

Overview of Changes in audiowmark-0.6.1:

• improve speed detection/correction
  • performance optimizations to make --detect-speed faster
  • improve accuracy of speed detection
  • make it work properly with short payload
  • add second, slower / more accurate algorithm (--detect-speed-patient)
• fix segfaults during hls-prepare (#11)
• read all input if a process provides audio on stdin to avoid SIGPIPE (#19)
• improve infrastructure for testing audiowmark
  • run some scripts for make check to ensure everything works correctly
  • add CI which builds/tests audiowmark automatically using github actions
  • support various sanitizer builds / STL C++ debug builds
  • fix some issues triggered by sanitizers
• add --strict option: provide strict and more permissive mode
  • input frames != output frames is only an error if --strict is used
  • enforce payload size if --strict is used
• improve command line parsing error messages
• documentation updates
• minor fixes