Converting 5.1 to 2-channel AAC

[ikarius]

So, I'm doing conversions from my DVDs and Blu-rays to .m4v x264 files. I'm using Plex for managing my video library, and my primary display client is a Roku box. I'm using AnyDVD-HD + Handbrake to do my conversions.

I've got a few questions about audio. The Roku is only capable of sending 5.1 audio out if it's in Dolby AC3 format. It's also capable of directly handling AAC, but only 2-channel formats.

When I set up the audio tracks to be put into the video files, I am planning to put in one 5.1 AC3 track and one 2-channel AAC track, but I've got some questions about the encoding options;

1. I can select multiple bitrates and sampling rates for the ffmpeg encoding of the 5.1 AC3. Almost all audio tracks for AC3 seem to be 48khz, and the handbrake guide says that AC3 bitrate is normally 448 kb/s. If I drop the sample rate to 44.1 and/or the bitrate to 384, am I setting myself up for compatibility issues where devices may not like those rates down the road?

2. On the AAC side, a Roku can only manage 2-channel output from an AAC stream, despite the fact that there are modern 6-channel discrete variants of AAC. The AAC encoding options include Dolby Prologic II, amongst others. I was under the impression that Prologic II was purely a post-processing effect which could upconvert 2-channels to 5.1 algorithmically. The handbrake guide really doesn't explain what it means to choose this over a stereo encoding- can anyone explain to me what the effect of choosing Prologic II encoding would be?


Thanks
Ikarius

[jman98]

The reason AC3 is at 48 kHz is that DVDs use a 48 kHz sample rate, as does BluRay (however, BluRay can accept higher sample rates for AC3). The first digital disc format, audio CDs, uses a 44.1 kHz sample rate because of limitations of late 1970s/early 1980s technology at the time it was invented. DVDs require a 48 kHz sample rate. You gain absolutely nothing by going to 44.1 kHz and someone could argue that you'll actually make the quality somewhat worse. 448 kbps is used because this is highest allowed bit rate for AC3 on DVD (BluRay allows higher). You can use a lower bit rate, but you'll be re-encoding and lessening the quality, even if you can't tell. But don't use 44.1 kHz. There's NO reason at all short of insanity for that. You could argue about saving space with lower bit rates, but lower sample rates won't save you any space.

Must you use AAC? AC3 is well supported and will be for years to come. Converting AC3 to AAC gains you nothing and any such conversions are lossy, even if you can't notice it. I don't deal with Handbrake and I avoid AAC where possible, so I have no idea what the Prologic II setting does.

[ikarius]

Thanks for the info on the sample rates. I'm including the AAC track in order to hedge my bets on compatibility, especially for options like mobile devices; Plex can also stream to android devices, and I've got a nexus 7 tablet, so I figure having the AAC track there is a backup plan. Hopefully someone can illuminate a bit about what the AAC encoding options mean.

[Island_Dweller]

Do your DVDs not have a 2.0 ac3 track? Most of my DVDs do.

Why not use the stereo ac3 to make a stereo aac?

[hello_hello]

The general "rule" seems to be 192kb/s for stereo AC3 and 448 kb/s or more for 5.1ch, but of course you can use lower bitrates if you want to reduce the file size. I agree, converting from 48k to 44.1k is fairly pointless. Then again, I also think converting AC3 audio in general is pretty pointless these days. Unless file size is an issue or for some reason or you need stereo and the original video only contains 5.1ch audio. Converting won't reduce the file size by a huge amount unless you want to drop the bitrate by a fair bit, so I just copy (use the original) AC3 audio. If HandBrake can't do it (although I think it can) I'd use something else.

DTS is a different story if it uses the full bitrate (there's generally 2 bitrates used for DTS on discs). The first is a bit over 700kb/s while the second is a bit over 1500kb/s (I can't remember exactly). If I do convert the audio to AAC I convert it using the default quality setting of 0.50 (I use MeGUI which uses the Nero AAC encoder). That gives you a variable bitrate as encoding video does, or as MP3s often do, and the same quality each time. I don't know if HandBrake has a quality method for AAC encoding.
Using the default quality setting I don't bother re-encoding AC3 or low bitrate DTS as it doesn't reduce the file size enough for me (unless you downmix 5.1ch to stereo), so I just copy those. For high bitrate DTS is can reduce the audio file size when converting from 5.1ch DTS to 5.1ch AAC by around 700MB for a movie.

The Dolby Prologic II method does what you described, it re-encodes the 5.1ch as stereo using Dolby Prologic II so it can be decoded as surround sound again using a Dolby Prologic II decoder. It's not a post processing effect which "fakes it" as such, it does use real multi channel audio as it's source. Encoder GUIs will usually do the same for other audio formats such as MP3. It won't sound as good as discrete 5.1ch audio though and as I'd never decode it using Dolby Prologic II myself I don't bother with it, but played back in stereo it should sound the same as a straight 5.1ch to stereo downmix.
And of course when you downmix using Dolby Prologic II you're re-compessing the audio so I assume lower bitrates will encode the surround information less accurately. In fact I don't really know how much of the surround sound information is lost when compressing it with a lossy encoder even at high bitrates. I've never listened to it decoded as Dolby Prologic II to find out.

[ikarius]

Well, the first question about converting to AC3 is simple; Roku can only pass a true 5.1 signal if it's encoded in AC3 format. Beyond that, AC3 is probably the best supported 5.1 channel format, so it's not much of a stretch to go with that for the higher quality- as long as it's supported. I do expect that any device in the future which I'd use as a plex endpoint in my home theater system would support AC3.

AAC does look like the format of the future, and having a 2-channel AAC track may be useful for streaming to mobile devices, such as an android tablet. I was asking about what the prologic II does, because I always thought it was generating 5.1 channels of info via postprocessing stereo info, rather than it being an actual encoding format; looking at info like here: http://www.axiomaudio.com/dolbyprologicII bolsters that thinking. I finally found today that PLII also understands 90-degree phase shifting for surround channels, so it is both a post-processing algorithm and an encoding format. Before finding that, I was really confused about what PLII could actually mean for the encoding.