View Full Version : Downmix quality on Standalones?

25th July 2003, 13:10
Hello all. I am currently in the process of doing a 9gig -> 4.7 gig dvd conversion, and the bitrate is alarmingly low (~3000K) with all the audio streams left in. Each track has 5 audio streams; a 2.0 and 5.1 in japanese and english, plus a 5.1 track with just the score. The 2.0 tracks are ~192K each, and there is a noticeable improvement to the video stream with the extra ~400K that gives me.

I do most of my playback here at college on my computer, but the end target is a settop. Unfornatly, right now I do not have a settop that likes DVD-RW's, so testing is expensive. My question is essentially this. How good is the 5.1 -> 2.0 conversion in most settop players? The quality gain in the video stream is noteciable but minor, but if it causes a noteciable audio quality loss cause of the downmix, it is definatly not worth it.


P.S.: Also, how much better do receivers do downmixing? Theoritically, should the output be identicle to the 2.0 stream after a proper downmix i.e. do the dvd authors usually reathor the 2.0 stream, or do they usually use a very advanced downmixer on the 5.1 track?

26th July 2003, 13:52
A surprisingly hard question to answer, because it is so subjective.

In my experience with SVCD conversion, there is a noticeable difference between 2.0 downmixed audio and the original 5.1 AC3. Of course with SVCD, there's also a conversion to MP2 in the mix, which gives lower audio quality over and above the 5.1-to-2.0 downmix.

For most people, I'd say you wouldn't notice the difference unless you were trying to notice the difference (or if you did a side-by-side A/B comparison). In other words, for casual everyday use a 2.0 downmixed stream would be fine.

However, for your purposes, I don't think you need to downmix. Your DVD already contains 2.0 tracks, which undoubtedly are already as good as or better than a downmix would be. And it lets you answer this question for yourself - put the original DVD in your settop player, switch between the 5.1 and 2.0 tracks and see how much of a difference you can hear.

If the difference is acceptable to you, just replace the 5.1 audio streams with dummy audio and you gain two whole AC3's worth of bitrate (something like 750 kbps IIRC - 25% more bitrate; nothing to sneeze at).

26th July 2003, 14:11
PS. I expect the downmixing implementations in DVD players and in receivers use algorythms similar (maybe identical) to the software implementations. However, because the players and receivers have dedicated firmware to do the processing, I expect they have additional stages to massage and fine tune the downmixing. Dedicated firmware almost always does a better job than software - the only advantage software has is that it doesn't always have to operate at real-time speeds.

I also expect that studios do it both ways - create 2.0 tracks from their original source recording, and create 2.0 tracks by downmixing the 5.1 track. But, I think it is safe to say (with very rare exception) that their results will always be as good as or better than our results. At best, they'll create the 2.0 surround stream from the original multi-track recordings. At better (less than best), even if they do downmix the AC3 5.1 to a 2.0 stream, they'll use dedicated hardware to do it. At worst, they'll downmix with their own software and get nearly the same results we do (with the rare exception here that they use inferior software - the open source community really does often produce the best quality software).