ffdshow audio options explained anywhere?

[shae]

Is there anywhere an explanation for ffdshow's audio options? In particular the Mixer settings (voice control, expand stereo, normalize matrix, LFE checkbox).

[foxyshadis]

LFE means a separate channel for a subwoofer. So if you have a real 2.1 or 5.1 (not a 2.0 with low-freq crossover to the sub, has to be a line run directly from the card to the sub), you'd check that.

Voice control is really a 3dB center channel boost. Expand stereo shifts the stereo image away from the center, making the two sides more distinct, as if you put the speakers to your side instead of in front of you. Normalize means the volume shouldn't be boosted when the values in the matrix add up to 2.4 or something like that, it reduces everything evenly so they all add up to 1.0. (Otherwise, >1.0 would raise the total output volume, although real volume is highly dependent on how it's distributed around speakers.)

[shae]

LFE means a separate channel for a subwoofer. So if you have a real 2.1 or 5.1 (not a 2.0 with low-freq crossover to the sub, has to be a line run directly from the card to the sub), you'd check that.But the dropdown has specific options for x.1. This lead me to think the checkbox adds the LFE channel to the mixdown, but then again, it isn't greyed out also when selecting x.1... so I'm not sure.

Voice control is really a 3dB center channel boost.Which you can achieve by sliding the voice slider above 100%, no? (Although the slider would boost by a relative amount rather than absolute). Wouldn't it be better named "+3db voice boost", or just drop it and have only the percentage slider?

Expand stereo shifts the stereo image away from the center, making the two sides more distinct, as if you put the speakers to your side instead of in front of you.Wouldn't that be better off with a slider too?

Normalize means the volume shouldn't be boosted when the values in the matrix add up to 2.4 or something like that, it reduces everything evenly so they all add up to 1.0. (Otherwise, >1.0 would raise the total output volume, although real volume is highly dependent on how it's distributed around speakers.)What makes the 2.4 threshold special? And what is it good for actually (I could understand if the theshold was 1.0)?

(Apparently it lowers the volume also without a custom matrix. Shouldn't the default matrix(es) add up each channel to 1.0?)

[foxyshadis]

The x/y in the dropdown means front/rear, not normal/lfe. That confused me for a long time, I guess that means it's a good idea to change it.

Sorry, 2.4 was just an example. There's no actual threshold, it's just the final sum of each speaker's inputs.

I have to look deeper into the mixer to find out exactly what each channel adds up to normally, heh. As well as whether the voice control and voice slider really control the same thing. It didn't look like it earlier, and if not, that'd be baaaaaaad naming. And yes, a slider would be better for the stereo boost, but I don't think the current algorithm is capable of handling that; it's currently simple algebra, but an fft is needed for handling arbitrary phase changes. On the other hand, phaeron put a good one in virtualdub.

[shae]

Ah, yes, "front/rear" was unexpected (and I did wonder how common a 3.2 configuration would be... ;)

As for a stereo expansion slider, I assumed the current algorithm just multiplied by a constant factor which could be changed with a slider. If it requires more intensive computation, perhaps it's not worth it; the extra cycles might be better spent elsewhere.