[FFmpeg-devel] [PATCH 1/3] avcodec: add siren audio decoder

[Lynne]

Feb 19, 2020, 17:58 by jamrial at gmail.com:

>>>> +AVCodec ff_siren_decoder = {
>>>> +    .name           = "siren",
>>>> +    .long_name      = NULL_IF_CONFIG_SMALL("Siren"),
>>>> +    .priv_data_size = sizeof(SirenContext),
>>>> +    .type           = AVMEDIA_TYPE_AUDIO,
>>>> +    .id             = AV_CODEC_ID_SIREN,
>>>> +    .init           = siren_init,
>>>> +    .close          = siren_close,
>>>> +    .decode         = siren_decode,
>>>>
>>>
>>> Missing a flush() function to clear s->backup_frame, s->imdct_in,
>>> s->imdct_out and s->imdct_prev.
>>>
>>
>> DO not like that.
>>

You kind of need to do that to support proper seeking.
You can keep s->imdct_in as is however since I don't see it used as anything other than per-frame storage.