[FFmpeg-devel] [PATCH] E-AC-3 spectral extension

[Loren Merritt]

On Tue, 2 Jun 2009, Michael Niedermayer wrote:
> On Mon, Jun 01, 2009 at 11:37:08PM -0400, Justin Ruggles wrote:
>> Michael Niedermayer wrote:
>>
>>> about the PRNG, well just decode a AC3 with 2 different PRNGS and compare
>>> by how much they differ
>>
>> LFG vs. MLFG
>> stddev:   14.99 PSNR: 72.80 bytes:  3999744/  3999744
>>
>> Nero vs. FFAC3
>> stddev:   19.75 PSNR: 70.41 bytes:  3998720/  3999744
>
> nice, so we are down to a stddev of ~4.76 that looks alot healthier already
> than the >100 of before, still its a little larger than what it should be,
> of course that might be a issue in nero as well instead of our decoder ...

Nitpick: assuming independence, it's variance that's additive, not stddev.
The amount of nero vs ffac3 difference that isn't explained by RNG is
sqrt(19.75**2 - 14.99**2) = 12.86

--Loren Merritt