WavPack 4.0 Beta Release 2, including sources Options

[bryant]

The second beta release of WavPack 4.0 includes the following:

• stable format (beta1 files okay too; beta1 decoder obsolete)
• decodes and plays legacy WavPack files (vers 1.0 to 3.97)
• sources available with new decoding library

download win32 executables plus winamp2/5 & foobar2000 plugins
download sources

This post has been edited by bryant: May 14 2004, 08:06

[den]

Ok, you asked for it...

kraftwerk1:
264 Kb WV hb256 - ABX 10/10
259 Kb OFSX b256 - ABX 10/10
In 2.4 - 3.4 seconds there are three tones (which I used to ABX Wavpack) and a "bwarp" type tone before. With the OFSX file, the edges of the "bwarp" sound are slightly different.

waiting:
270 Kb WV hb256 - ABX 8/10
256 Kb OFSX b256 - ABX 10/10
In 11.1 - 16.2 some of the edges to the noise are not quite the same. The slight difference I could hear around the vocals in hb256 Wavpack are not present here. They disappear with hb256x6 also if you read above.

Another one bites the dust:
259 Kb WV hb256 - ABX 10/10
265 Kb OFSX b256 - ABX 10/10
Slight changes in the noise in the percussion. "Dustiness" is not so noticeable here.

porcelain:
267 Kb WV hb256 - ABX 10/10
256 Kb OFSX b256 - ABX 10/10
Again, the noise in the samples that come in is not quite the same, but damn close.

rushing:
261 Kb WV hb256 - ABX 10/10
260 Kb OFSX b256 - ABX ?/10
Haven't played with this one too closely yet, but a quick listen sounds transparent.

OK, the above might read like a negative set of results, but it shouldn't be read this way. I had to really pore over each sample to find differences to ABX, and it took more time to find something compared to the Wavpack equivalent hb256 samples. I haven't conducted a close comparison between OFSX and Wavpack 4, but the OFSX samples were generally "closer to transparency". I achieved the 10/10 ABX results more because I knew what to listen for after playing with Wavpack 4. If hb256x6 is used with Wavpack 4 instead, there is still some difference between the two, but the gap is often closer.

The "artifacts" I'm hearing are barely there, and are best described as minor changes in noise in the background in most cases. If you were to listen to the above OFSX versions by themselves, you would probably miss it. It only becomes vaguely noticeable in a focussed ABX situation with short samples.

If I had to pick one of these encoders at this bitrate based on these samples, purely for quality it would be OFSX, but we don't know how long these samples took to encode. If Wavpack is given plenty of time to encode (ie hb256x6) it gets quite close.

I wouldn't trash your encoder Florin.

I also think that both you and David need to be congratulated with these latest efforts. At 256ish kbit, these encoders are now seriously in the same league as the other popular lossy encoders.

Hope this helps.

Den.