Wavosaur doesn't host latency. So all WAV files processed through it with Stereo Tool will be shifted forwards 4096 samples, or 6976 samples if Delossifier is enabled. If there isn't at least that many samples of silence at the end of the WAVs, they will get cut off. (6976 samples is about .158 seconds @ 44,100 hz).
I have seen it, where part of a previous WAV file will appear in the blank 4096/6976 samples at the beginning of a processed wav. This actually also happens in older versions of Audacity back when it didn't host the latency.
You can use Sox
to "pad" your files with 4096 or 6976 samples at the end of your files, then when done processing, to "trim" the 4096 or 6976 samples off from the beginning.
To pad files, put this to "pad.bat" batch file, then run it in the directory with your wavs (with sox.exe, or put sox.exe into a PATH environment variable on your system):
for %%i in (*.wav) do sox.exe "%%i" "pad\%%~ni.wav" pad 0 4096s
And here's the batch file for trimming:
for %%i in (*.wav) do sox.exe "%%i" "trim\%%~ni.wav" trim 4096s
Keep in mind that while Sox supports 32 bit float wavs, its internal
code only supports 32 bit pcm. So, if you process any 32 bit float wavs through it that have clipping (audio levels above 1.000), Sox will
permanently clip the files at 1.000. You can batch check the levels of your files using Foobar. (Add files, right click all selected files, "Replay gain" -> "Scan per-file track gain".)