WavPack

WavPack is an open audio compression format developed by David Bryant, usually using the `.wv` extension. It can operate in pure lossless, lossy, and hybrid modes and supports integer PCM, floating-point PCM, multichannel audio, and DSD data in later versions.

In lossless mode, WavPack reduces audio redundancy through prediction, channel decorrelation, and residual encoding, restoring original samples on decode. Encoding options change prediction complexity, processing speed, and file size but should not change correctly recovered data. The format organizes audio and metadata in independently parseable blocks, aiding streaming, error localization, and multichannel extension.

Hybrid mode is one of WavPack's distinctive operating methods. The encoder produces a standalone playable lossy `.wv` main file and may also generate a `.wvc` correction file; together they let the decoder restore lossless audio identical to the encoding input. A `.wvc` file is not a second complete standalone track; if the correction file is missing, the main file still plays but only as lossy output.

WavPack's lossy mode is not identical in design to MP3 or AAC. It reduces data by quantizing prediction residuals and related methods, emphasizing a structure suitable for later combination with a correction file; perceived quality and compression ratio still depend on settings and source material. Converting a standalone lossy main file to an ordinary lossless format cannot replace a missing `.wvc`.

Tags usually use the APEv2 structure and can store text, cover art, and replay gain. WavPack can also encapsulate certain non-audio supplementary information to rebuild the original file structure on decode; "restore original file" and "restore audio samples" should be judged separately according to encoding options and input type.

WavPack, FLAC, and ALAC can all store PCM losslessly, but bitstreams, metadata, hardware support, and encoding characteristics differ. The extension indicates expected format only; verifying lossless archives still requires decode verification, sample comparison, or in-format checksum mechanisms, not file size alone.