[opus] Formula/heuristic for estimating packet size?

[Albin Stigö]

Ok. Will OPUS_GET_BITRATE reflect the actual bitrate if packet size is
too small?

--Albin

On Sat, May 12, 2018 at 7:30 PM, Jean-Marc Valin <jmvalin at jmvalin.ca> wrote:
> Note also that the packet size you give the encoder also acts as an
> absolute max on the bitrate. For example, if you ask for 32 kb/s VBR but
> give a max packet size of 120 bytes, then you're absolutely certain the
> bitrate will never go over 48 kb/s.
>
>         Jean-Marc
>
> On 05/12/2018 12:42 PM, Albin Stigö wrote:
>> Just a follow up... I guess I was a bit confused about the VBR
>> setting. I realise now that packets tend to stay very close to
>> OPUS_SET_BITRATE so that solves my problem.
>>
>> --Albin
>>
>> On Sat, May 12, 2018 at 6:19 PM, Albin Stigö <albin.stigo at gmail.com> wrote:
>>> Thanks for the input!
>>>
>>> --Albin
>>>
>>> On Sat, May 12, 2018 at 6:00 PM, Orestes Zoupanos
>>> <oresteszoupanos at hotmail.com> wrote:
>>>> Hi Albin!
>>>>
>>>> There may be some details at: https://tools.ietf.org/html/rfc7845#section-6
>>>>
>>>> Otherwise, I hope someone else on this list might be able to give a better
>>>> formula for estimating packet size.
>>>>
>>>> Regards,
>>>>
>>>> Orestes
>>>>
>>>> On 12/05/2018 16:31, Albin Stigö wrote:
>>>>
>>>> Hello,
>>>>
>>>> I'm working on an Opus encoder block for GNUradio (a signal processing
>>>> toolkit).
>>>>
>>>> I was wondering if there's some formula/heuristic for estimating the
>>>> packet size average case / worst case given a certain encoder setting
>>>> (assuming VBR).
>>>>
>>>> I need to provided a reasonable estimate to the GNUradio memory allocator.
>>>>
>>>>
>>>> --Albin
>>>> _______________________________________________
>>>> opus mailing list
>>>> opus at xiph.org
>>>> http://lists.xiph.org/mailman/listinfo/opus
>>>>
>>>>
>>>> _______________________________________________
>>>> opus mailing list
>>>> opus at xiph.org
>>>> http://lists.xiph.org/mailman/listinfo/opus
>>>>
>> _______________________________________________
>> opus mailing list
>> opus at xiph.org
>> http://lists.xiph.org/mailman/listinfo/opus
>>