2011/4/6 Aℓex Converse <[email protected]>:
> On Tue, Apr 5, 2011 at 4:41 PM, Nathan Caldwell <[email protected]> wrote:
>>
>> Update attached.
>>
>
> This looks good at a 10000 foot level. I expect to have some time to
> do a through review friday or this weekend.
This chunk seems kind of funny to me:
+ minsnr = pow(2.0f, pe_min / band_sizes[g]) - 1.5f;
+ /* Limit min SNR between -25dB and -1dB */
+ coeff->min_snr = av_clipf(1.0f / minsnr, 3.1622776e-3f,
7.9432821e-1f);
[...]
+static float calc_reduced_thr_3gpp(AacPsyBand *band, float min_snr,
+ float reduction)
+{
+ float thr = band->thr;
+
+ if (band->energy > thr) {
+ thr = powf(thr, 0.25f) + reduction;
+ thr = powf(thr, 4.0f);
+ if (thr > band->energy * min_snr && band->avoid_holes !=
PSY_3GPP_AH_NONE) {
+ thr = FFMAX(band->thr, band->energy * min_snr);
+ band->avoid_holes = PSY_3GPP_AH_ACTIVE;
+ }
+ }
+
+ return thr;
+}
Please include a note about deviating from the spec.
+ /* Limit min SNR between -25dB and -1dB */
+ coeff->min_snr = av_clipf(1.0f / minsnr, 3.1622776e-3f,
7.9432821e-1f);
Please make these defines, the latter value is used all over the patch.
Otherwise Ok
--Alex
Otherwise OK.
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
