[sip-comm-dev] JavaEncoder/Decoder for celt


#1

Hi Lubo, Jean and Emil,

I have finished writing and testing of the JavaEncoder and JavaDecoder. With
them, I encode a .raw audio file and then use JavaDecoder to decode the
compressed file back to .raw file. Again it sounds fine.

Because i am not quite familiar with the JMF/FMJ usage in SIP( i can't find
an application example of using JMF codec, but can only refer to the API
documentation and make some assumptions of how to use it), there may be
some misunderstanding of FMJ in the code. I have written detailed comments
about my understanding of FMJ and the assumptions about the usage in the
source code (see the attachment), could you please help me review it and
point out my misunderstanding. Thanks in advance.

And another problem I think I should report to you. I did another test
"tandom" these two days, which do the encode-decode and feed the output of
decoder back to encoder, all these again and again for several loops, to
see if the final output increase or decrease too much in energy. And there
are still some results not pass the threshold suggested by the author of
celt. I don't know how important is this (confirm this for me?), but I'll
try to see more about it.

The next two days I will be busy moving to my new dormitory, so I may have
less time to do this, and may reply a little slowly. I will try to fix these
problems once everything settled down.

Regards,
Jing

JavaEncoder.java (16.5 KB)

JavaDecoder.java (14.6 KB)


#2

I have finished writing and testing of the JavaEncoder and JavaDecoder. With
them, I encode a .raw audio file and then use JavaDecoder to decode the
compressed file back to .raw file. Again it sounds fine.

It was already said in a chat conversation about a week ago but it's
being put there for the sake of the completeness of this thread: the
offered JMF Codec classes for the CELT encoder and decoder are
considered sufficient for now and the focus of the development is to
be moved to the second codec for Google Summer of Code 2010.

Because i am not quite familiar with the JMF/FMJ usage in SIP( i can't find
an application example of using JMF codec, but can only refer to the API
documentation and make some assumptions of how to use it),

SIP Communicator is the definitive example of using JMF codecs.

there may be
some misunderstanding of FMJ in the code. I have written detailed comments
about my understanding of FMJ and the assumptions about the usage in the
source code (see the attachment), could you please help me review it and
point out my misunderstanding. Thanks in advance.

As I've already written on the dev mailing list before, we're not
running on top of FMJ and our only concern during the development of
Codec implementations is JMF.

And another problem I think I should report to you. I did another test
"tandom" these two days, which do the encode-decode and feed the output of
decoder back to encoder, all these again and again for several loops, to
see if the final output increase or decrease too much in energy. And there
are still some results not pass the threshold suggested by the author of
celt. I don't know how important is this (confirm this for me?), but I'll
try to see more about it.

Let's first finish the second codec because it's of higher priority
and leave the finishing touches to CELT for the end of Google Summer
of Code 2010.

ยทยทยท

On Sun, Jul 18, 2010 at 7:10 AM, Dai Jing <daijing09@gmail.com> wrote:

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@sip-communicator.dev.java.net
For additional commands, e-mail: dev-help@sip-communicator.dev.java.net