WavPack Audio
WavPack can compress audio losslessly or use lossy and hybrid modes with an optional correction file.
You have a small WV music file and a larger matching WVC file. Playing the WV works by itself. WavPack hybrid mode deliberately provides a lossy stand-alone file plus correction information for lossless restoration.
WavPack also has ordinary lossless and lossy modes, so identifying the actual arrangement is the first step.
| Arrangement used by the decoder | Result |
|---|---|
| Lossless-mode WV | Restored supported source data |
| Lossy/hybrid WV alone | Its lossy reconstruction |
| Hybrid WV plus matching WVC, used by a capable decoder | Lossless restoration of that source |
Why the correction file changes the result
An illustrative source sequence is 1,000, 1,001, 1,004. Suppose the stand-alone representation reconstructs 1,000, 1,000, 1,008. Correction values of 0, +1, and −4 supply the missing differences.
| Stand-alone reconstruction | Exact correction | Combined result |
|---|---|---|
| 1,000 | 0 | 1,000 |
| 1,000 | +1 | 1,001 |
| 1,008 | −4 | 1,004 |
This is a model of exact correction, not WavPack's literal packet coding. It explains why the pair can restore data that the WV alone cannot identify.
A player must actually locate, validate, and use the matching correction information. A WVC kept in an unrelated folder or ignored by the decoder contributes nothing to that listening result.
Prediction and noise management perform different jobs
WavPack uses reversible prediction/decorrelation and compact coding for lossless operation. The exact residual information is sufficient to undo those transformations.
Hybrid/lossy operation makes a controlled approximation in the independently playable part. Noise-shaping choices can affect that part's error spectrum. The correction information restores what is needed for the lossless result rather than asking the listener to infer it from the approximation.
The two modes therefore answer different storage/delivery needs without requiring the source to be separately encoded into two unrelated music files.
Preserve the actual sample representation
WavPack can store integer PCM, floating-point PCM, and losslessly compressed DSD. DSD support uses lossless storage rather than the PCM hybrid arrangement.
Those distinctions matter when converting. A target supporting the same integer data can preserve it. A target designed for integer PCM cannot automatically preserve every float value or a DSD bit sequence unchanged. Converting DSD to PCM filters and changes representation; unpacking lossless DSD restores its original one-bit data.
WavPack can also retain wrapper information for restoration, depending on the operation and settings. Audio equality, original-file reconstruction, and tag preservation are separate verification goals.
For a WavPack item delivered through Emby, identify whether the decoder used a lossless WV or a hybrid pair, and what sample representation it reconstructed. That establishes the source result before any subsequent delivery or output conversion.