On Fri, Jul 29, 2011 at 1:58 PM, Nathan Caldwell <[email protected]> wrote:
> The spec states:
>
> * Only the lowest 12 spectral coefficients of any LFE may be non-zero
>
> We were using the 12 lowest *bands*.
> ---
> libavcodec/aacenc.c | 10 +++++++++-
> 1 files changed, 9 insertions(+), 1 deletions(-)
>
> diff --git a/libavcodec/aacenc.c b/libavcodec/aacenc.c
> index e74af0b..7b3794f 100644
> --- a/libavcodec/aacenc.c
> +++ b/libavcodec/aacenc.c
> @@ -540,6 +540,14 @@ static int aac_encode_frame(AVCodecContext *avctx,
> wi[ch].window_shape = 0;
> wi[ch].num_windows = 1;
> wi[ch].grouping[0] = 1;
> +
> + /* Only the lowest 12 coefficients are used in a LFE channel
> */
> + if (s->sample_rate_index >= 11)
> + ics->num_swb = 1;
> + else if (s->sample_rate_index >= 8)
> + ics->num_swb = 2; /* This actually results in 16
> coefficients */
Should we play it safe and only use 8 here? Can we zero out 13-16
after the fact?
> + else
> + ics->num_swb = 3;
> } else {
> wi[ch] = s->psy.model->window(&s->psy, samples2, la,
> cur_channel,
> ics->window_sequence[0]);
> @@ -550,7 +558,7 @@ static int aac_encode_frame(AVCodecContext *avctx,
> ics->use_kb_window[0] = wi[ch].window_shape;
> ics->num_windows = wi[ch].num_windows;
> ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
> - ics->num_swb = tag == TYPE_LFE ? 12 :
> s->psy.num_bands[ics->num_windows == 8];
> + ics->num_swb = tag == TYPE_LFE ? ics->num_swb :
> s->psy.num_bands[ics->num_windows == 8];
> for (w = 0; w < ics->num_windows; w++)
> ics->group_len[w] = wi[ch].grouping[w];
>
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
