[Ffmpeg-devel] AAC decoder

[Benjamin Larsson]

Mike Melanson wrote:

> Benjamin Larsson wrote:
>
>> And the sine_short table should be a subset of sine_long.
>
>
>     That seems intuitive, but I don't think it's true in this case.
> The short values are derived as:
>
>   for odd values i in 1..127
>     sine_value = sin(i * PI / 512)
>
> For long values:
>
>   for odd values in 1..1023
>     sine_value = sin(i * PI / 4096)
>
> In the short formula, you could multiply (i * PI / 512) by (8/8) to
> get the same denominator as in the long formula. But multiplying by 8
> would make the numerator even and thus the sine value would not occur
> in the long table.

You are right, the term (that you missed here) makes it impossible, but
I think that it would be possible to approximate the window by tweeking
the term abit to reduce the error. And you got the term wrong, it should
be 0.5.

>
>     (I think.)

MvH
Benjamin Larsson

-- 
"incorrect information" is an oxymoron. Information is, by definition, factual, correct.