I had never heard of the lossless compressor LiteWave that avibe found in the thread about lossless ACMs, so I decided to give it a whirl. The command-line version is free, but the ACM codec is not (and no price is given on the website), so I didn't test that part. It it a Win32 program, has lossless and lossy modes, and I found to be easy to use (straightforward syntax, nice help, and it didn't choke or gag on anything I gave it). It would not handle 24-bit files.

I verified that it is truly lossless in the lossless mode. It has only a single speed/quality mode which runs a little slower than WavPack's "high" mode and compresses a little worse than WavPack's "very fast" mode. The lossy mode has a "q" parameter that sets the quality and can be set from 0 (lowest quality) to 100 (lossless). I determined that this works by simply selecting a quantization level for the whole file of +/-(1000 - 10q), so q=0 gives +/-1000 and q=99 gives +/10. Because a single quantization level is used throughout the whole file and no dither is applied, this mode sounds pretty bad except at the very highest settings.

In summary, the lossless and lossy modes really do not compete too well with existing compressors, but the ACM feature may make this attractive in some cases (assuming it's not $200 ).

Yeah, you'd think they'd at least put the cost on their webpage but I didn't see it.

I saw this Litewave a while ago and couldn't believe they were trying to make money out of it (given the existance of open source + other free alternatives which are in all respects except for the ACM feature superior) but good luck to them. It does raise the question though why no other lossless audio program as an ACM codec. Are they really that hard to code? (I notice now this was discussed in http://www.hydrogenaudio.org/forums/index.php?showtopic=1272 however I didn't see many answers to the question in this thread).

I had a look into writing an ACM codec myself once but couldn't really find enough information to get started (and the DDKs are pretty huge to download), however I'd be keen to write one if someone could point me in the right direction. What I think it would be really nice would be to find/create a nice framework which all the coders could then relatively quickly adapt to. Writing ACM / Winamp3 / whatever plugins are the type of thing I thing could+should be made so simple but are too often so annoyingly difficult.

Hey Michael!

My knowledge of this is pretty hazy, but it does seem to be difficult to adapt to existing formats.

First of all, the data must be embedded in a regular RIFF WAV file with a special format ID, so right away you have to have versions of your files with and without wav headers (kind of like mp3s sometimes have).

Also, the ACM acts as a filter, so the calling program just passes the data through the filter in chunks. It's possible (I think) that it might not even send the beginning of the data through the filter but might just start in the middle somewhere and expect you to synch up and figure everything out. And can it skip to another section without warning you? It may be that this LiteWave has such poor compression because it has all sorts of synch information and error checking embedded in the data so that it can work smoothly.

I think that ACM was designed for things like 4-bit ADPCM (CBR) and it's an enormous hack to get it to work reliably with a complex VBR format. Maybe we can get them to post the source of the LiteWave ACM somewhere for us to take a look!

Thanks - that explains why no ones implemented it then. So much for that idea .

Isn't there a LAME ACM codec? At least I think I remember seeing one... Perhaps it can give some useful info by looking at the source...