[FFmpeg-devel] [PATCH v2 9/9] avformat/pcm: decrease target audio frame per sec to 10

Tobias Rapp t.rapp at noa-archive.com
Fri Mar 15 09:54:15 EET 2024


On 14/03/2024 23:04, Marton Balint wrote:

>
>
> On Thu, 14 Mar 2024, Tobias Rapp wrote:
>
>> On 08/03/2024 01:53, Marton Balint wrote:
>>
>>>  This makes the wav and pcm demuxer demux bigger packets, which is more
>>>  efficient.
>>>
>
> [...]
>
>>>  +0,     136000,     136000, 2000,    24000,
>>>  f952be1e1c7d48acd73d652ef4ea3aad
>>>  +0,     138000,     138000,     2000,    24000,
>>>  e42cbabef9d9675429dcb6fa28ce39f4
>>>  +0,     140000,     140000,     2000,    24000,
>>>  9c6613b980bcbe9ebacbb6a492f53b16
>>>  +0,     142000,     142000,     2000,    24000,
>>>  1882d1ae4b08fa7cc848eba60d645463
>>>  +0,     144000,     144000,     2000,    24000,
>>>  a361b56ba31c90557689631f0e7afc63
>>>  +0,     146000,     146000,     2000,    24000,
>>>  341fbe24112c4060bb643750895a9041
>>>  +0,     148000,     148000,     2000,    24000,
>>>  38340ef2ef764bb8007efe9cd0af54d7
>>>  +0,     150000,     150000,     2000,    24000,
>>>  9b013a00bbedad211ba33326a3cfaa6e
>>>  +0,     152000,     152000,     2000,    24000,
>>>  9d3469da4442cdfd2b8a5beacb789701
>>>  +0,     154000,     154000,     2000,    24000,
>>>  cf8e3880ecb46011328135dba1015044
>>>  +0,     156000,     156000,     2000,    24000,
>>>  3be9fa7d0e0541269339ee032bda4021
>>>  +0,     158000,     158000,     2000,    24000,
>>>  ffd2646acf86524651416bc412afd846
>>>  +0,     160000,     160000,     2000,    24000,
>>>  715163d8f82f007b1b6c42b06bd08b5a
>>>  +0,     162000,     162000,     1840,    22080,
>>>  c2079a237d021d7aa27e44acf979b1f0
>>
>> Noob question: Isn't a much larger total number of samples produced 
>> here?
>
> The command line which generates the daud file specifies -frames:a 20, 
> and 20 from bigger packets will cause a longer output file.
>
Ah ok, I was just wondering because the packet count changes from 4 to 
65 while the packet size stays at 24000. Didn't notice that a two-step 
process (encoding with limited frame-count into file, then separate 
decoding to get the framemd5) is performed here in contrast to other 
tests with frame-count limit.
>>
>> Also it might be good to mention in the commit that the audio 
>> duration is less exact when using commands like "ffmpeg -i $INPUT 
>> -c:a copy -t 1.0 $OUTPUT".
>
> Ok, will add this to the commit message.
>
Thanks,
Tobias



More information about the ffmpeg-devel mailing list