On 12/18/2012 01:15 PM, Anton Khirnov wrote: > > On Sun, 16 Dec 2012 16:27:29 -0500, Justin Ruggles <justin.rugg...@gmail.com> > wrote: [...] >> + >> + if (c->method == AV_RESAMPLE_DITHER_TRIANGULAR_NS) { >> + if (sample_rate == 48000) { >> + c->ns_coef_b = ns_48_coef_b; >> + c->ns_coef_a = ns_48_coef_a; >> + } else { >> + c->ns_coef_b = ns_44_coef_b; >> + c->ns_coef_a = ns_44_coef_a; >> + } >> + } >> + >> + c->flt_data = ff_audio_data_alloc(channels, 1024, AV_SAMPLE_FMT_FLTP, >> + "dither flt buffer"); >> + if (!c->flt_data) >> + goto fail; >> + >> + /* Either s16 or s16p output format is allowed, but s16p is used >> + internally, so we need to use a temp buffer and interleave if the >> output >> + format is s16 */ >> + if (out_fmt != AV_SAMPLE_FMT_S16P) { >> + c->s16_data = ff_audio_data_alloc(channels, 1024, >> AV_SAMPLE_FMT_S16P, >> + "dither s16 buffer"); >> + if (!c->s16_data) >> + goto fail; > > Why are you allocating s16_data conditionally and flt_data always. > flt_data is only used when input format is packed, right?
Only when the input is not FLTP. I just forgot to change that since the previous patch which sometimes copied the input in order to modify it. I'll fix that, plus the other 2 things you mentioned. Thanks, Justin _______________________________________________ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel