2011/5/16 Måns Rullgård <[email protected]>

> Young Han Lee <[email protected]> writes:
>
> >
> > The buffer for ltp (ltp_state) is defined in "short" type in the
> reference
> > software and the document.
>
> Reference software is awful more often than not, and AAC decoding isn't
> precisely specified, so being more accurate than the reference code is
> allowed.
>
> > Of course "float" type does not make serious error but just rounding
> error.
>
> A float can represent all values an int16_t can, so this change can only
> make it more accurate.
>
>
I thought ltp was calculated using "short" in the encoder but it was not.
I totally agree with your opinion.

Regards,
Young Han Lee
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to