[FFmpeg-devel] [PATCH] lavc/audiotoolboxenc: fix noise in encoded audio

[wm4]

On Tue, 2 Jan 2018 22:27:49 +0800
Jiejun Zhang <zhangjiejun1992 at gmail.com> wrote:

> On Tue, Jan 2, 2018 at 8:03 PM, Carl Eugen Hoyos <ceffmpeg at gmail.com> wrote:
> > 2018-01-02 8:52 GMT+01:00  <zhangjiejun1992 at gmail.com>:
> >  
> >> @@ -565,6 +579,10 @@ static av_cold int ffat_close_encoder(AVCodecContext *avctx)
> >>      ff_bufqueue_discard_all(&at->frame_queue);
> >>      ff_bufqueue_discard_all(&at->used_frame_queue);
> >>      ff_af_queue_close(&at->afq);
> >> +    if (at->audio_data_buf_size > 0) {
> >> +        at->audio_data_buf_size = 0;
> >> +        av_free(at->audio_data_buf);
> >> +    }  
> >
> > The if() looks unnecessary.  
> 
> Yes. I'll remove it. Thanks for pointing it out.
> 
> > Could you explain what this patch changes?
> > From a quick look, until now a buffer a was used with a calculated size x.
> > After the patch, a buffer b with the same calculated size x is allocated,
> > and x bytes are copied from a to b.
> > What do I miss?  
> 
> Although undocumented, AudioToolbox seems to require the data supplied
> by the callback (i.e. ffat_encode_callback) being unchanged until the
> next time the callback is called. In the old implementation, the
> AVBuffer backing the frame is recycled after the frame is freed, and
> somebody else (maybe the decoder) will write into the AVBuffer and
> change the data. AudioToolbox then encodes some wrong data and noise
> is produced. Copying the data to a separate buffer solves this
> problem.

This should be in the commit message.