NEWS

Overview of Changes in audiowmark-0.6.4:

* reduce memory usage:
  - for speed detection, especially when using multiple keys
  - for resampling of input files with sample rate != 44100
  - split long input files into smaller chunks (get --chunk-size option)
* optimize performance for speed detection / normal sync
* improve detection for libgcrypt, use pkg-config if available (#72)
* fix problems in make check (#72)

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)

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)

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

Overview of Changes in audiowmark-0.6.0:

* implement speed detection/correction (--detect-speed)
* Add '--json' CLI option for machine readable results.

Overview of Changes in audiowmark-0.5.0:

* support HTTP Live Streaming for audio/video streaming
* fix floating point wav input
* improve command line option handling (ArgParser)
* support seeking on internal watermark state

Overview of Changes in audiowmark-0.4.2:

* compile fixes for g++-9 and clang++-10
* add experimental support for short payload

Overview of Changes in audiowmark-0.4.1:

* initial public release

Overview of Changes in audiowmark-0.4.0:

* add initial video watermarking support (videowmark)

Overview of Changes in audiowmark-0.3.0:

* replace padding at start with a partial B block
* add algorithm for decoding the watermark from short clips

Overview of Changes in audiowmark-0.2.1:

* add limiter to avoid clipping during watermark generation

Overview of Changes in audiowmark-0.2.0:

* support input/output streams
* support raw streams
* some performance optimizations
* unified logging and --quiet option
* improved mp3 detection to avoid false positives
* split up watermarking source (wmadd/wmget/wmcommon)

Overview of Changes in audiowmark-0.1.0:

* initial release