I certainly appreciate everyone's comments.
I have 16-bit, 8-bit, mono, and stereo tested at many differrent sample rates so far, but
it
should be able to handle any number of channels < 65536. I will work to make it
flexible enough to
handle 24-bit, 32-bit, and 64-bit, but no promises. =;)
I have no plans to learn how to use any sort of container format yet because I want to
concentrate
on compression and data types. Any container with frames can be used quite easily; just
shove
audio data encoded as JPEG 2000 JP2 images in a container using my methods; the
compression and
framing do not change. I will continue with the KISS approach and just modify my extremely
simple
j2a format for now. I will add some fields like sample rate, number of channels, number of
frames,
and precision to the header, so it can be read without decoding a frame.
I will also put an arbitrary header before the first audio frame to contain anything a
developer/user wants. This arbitrary header could be an id3 tag, photo, md5sum, j2a
file...anything.
fps
sample rate
number of channels
number of frames
precision
arbitrary header size
arbitrary header data
frame size
frame JP2 data
frame size
frame JP2 data
...
EOF
Someone pointed me to some great links and suggested that I test against additional
lossless
codecs, but I can only do that to a point because I must focus my efforts. The large
CD-quality
test data set compression numbers make me to want to test against the top 3 average
compression
ratio codecs: LA, OptimFROG, and Monkey's Audio. I will need to broaden my search once
I start
heavily testing 32-bit, but I am not there yet.
I would be happy to integrate j2kaudio into another library or program once I feel it is
mature
enough.
Thanks again for all of the helpful feedback. -Greg
--- Erik de Castro Lopo <mle+la(a)mega-nerd.com> wrote:
Thorsten Wilms wrote:
- Have
you thought of putting your codec data inside other standard
container formats like WAV, Caf and Ogg?
Or Matroska. Ogg and Matroska are not (yet?) supported by libsndfile,
though.
Unfortunately, Ogg seems to be a surprisingly difficult. To fit
easily into the libsndfile view of the world libsndfile needs
to be able to find out how many frames of audio are in the file,
preferably without doing too much seeking about.
So far I have only tried to do OggFlac and haven't managed to get
very far. I don't know if this is something specific to OggFlac or
more generally a property of Ogg. However, I do know that two
people have tried to add OggVorbis to libsndfile without being able
to complete the task.
Erik
--
-----------------------------------------------------------------
Erik de Castro Lopo
-----------------------------------------------------------------
"Imagine if every Thursday your shoes exploded if you tied them
the usual way. This happens to us all the time with computers,
and nobody thinks of complaining." -- Jef Raskin
_______________________________________________
Linux-audio-dev mailing list
Linux-audio-dev(a)lists.linuxaudio.org
http://lists.linuxaudio.org/mailman/listinfo.cgi/linux-audio-dev