[vorbis-dev] slightly different audio output

[Owen Emry]

Segher Boessenkool wrote:

>> I wonder if there is specific requirement or restriction for the decoder
>> in order to be consider "Vorbis" compliant (or let's say, to be a
>> "correct" vorbis decoder). I am experimenting with integerized libvorbis
>> and the decoder delivers the audio data as following, (and compare with
>> the floating point version below).
>
>First of all, there is no published formal spec yet.
>
>> $ tail audioout-int.txt
>> 0290670 025e 085c 02cf 0895 0349 08b2 039e 087d
>> 0290680 03ab 07df 03a0 0738 03de 06d6 0458 069f
>> 
>> $ tail audioout-float.txt
>> 0290670 0262 0860 02ee 08a1 0377 08c5 03bc 0884
>> 0290680 03c7 07ea 03be 073c 03d8 06c0 0459 0697
>
>But these differences seem way too big.
>
>> In theory, must they deliver exactly the same audio output or some
>> errors are tolerable.
>
>Some slight differences will be unavoidable, I think.  But nothing
>bigger than off-by-one.

Off by one what?  One in the LSB of a 16-bit integer sample?  Nonsense.  :)
Think about it:  Take an original piece of PCM data, encode it with Vorbis,
and then decode it again using the reference libvorbis.  Certainly you would
expect the vorbis step to inflict more than an "off-by-one" effect on each
sample.

Thus perhaps the real question is whether or not a given libvorbis delivers
results that are at least as "lossless" as the reference library.  And, as
another poster noted, you can't really tell just from looking at raw data.
It requires a more involved analysis...

-Owen

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.