I've no idea whether this will be welcomed as a good idea (as it could
drag in other developers and advance it's acceptance/use), rejected as
a bad idea, or even resented as a trespassing idea. :) But has
anyone ever considered offering or trying to integrate the workings of
Codec2 into a better known or/and more popular codec for use in
specific modes?
Ideas like this pop up on this list every now and then, most recently
earlier this year. But that was about coming up with a container if I
remember correctly, not smushing together two codecs
For instance right now Opus is pretty much the king of all lossy
codecs to my knowledge. It beats everyone at every bitrate across the
board without exception at least down to 8kbps or so and can be used
in various specialized uses such as realtime communication with low
latency, not only latency-irrelevant uses such as streaming media.
Although Codec2 is based on totally different principles, in my not so
humble opinion it is the king of low bitrate voice only codecs that
i've heard. The whole reason I joined the mailing list was due to
interest in the codec over that fact, having compared it to every
other super low bitrate codec i'd listened to including the standards
like MELPe.
It would not be unprecedented to have totally different compression
algorythms under 'special modes' for lower bitrate, I believe Windows
WMA9 did that having separate modes for lossless, lossy music, and
lossy voice. I do not know the rate of finalization of Opus but I do
know that if any kind of merging or enhancements would ever join the
project, it would make sense to have it happen earlier instead of
later when it's too late to be a part of the state of the art.
I'm not sure if you know, but Opus is already the combination of two
codes. Skype's SILK for the voice part (frequencies below 8 kHz) and
CELT for the "music" part (frequencies above 8 kHz). Opus is also quite
finalized as I understand it. I stopped following its development some
time ago, so I may be wrong
Alternately it could join some other audio codec if there is anything
on the horizon that looks to out-everything Opus, but I have not heard
of it. Unlike in the world of video where it's gone from h264 to h265
and VP9 to VP10, i'm not aware of any other "nextgen" audio codecs
that threaten to even match let alone dethrone Opus, though i'm all
ears if anyone is aware of them. Since Codec2 seems the best of the
best voice codecs even early in it's state to me it seems like a match
made in heaven - and i'll freely admit wanting to have the ability to
conveniently play Codec2 files on everything from computers to mobile
phones in the future, and maybe even dedicated hardware later.
I am aware Codec2's primary purpose is for realtime low bandwidth
communication - yet I see nothing against using it as a storage method
for space constrained primarily vocal media either, and i'd think the
development assistance could help advance things faster and propagate
use of it more widely, though now i'm getting redundant so i'll leave
it at that.
It'd certainly be useful for audiobooks and archives of political
speech, and podcasts
What say everyone, is this an idea so bad it should never be voiced
again, or something worthy of consideration?
I'd say it's a good goal to have, but there's plenty of important
details going into something like this. For example, seeking. You want
to use container which is sane and has a compact representation of seek
tables. This rules out AVI, Ogg and maybe MOV. Matroska is probably
fine. Being able to handle switching modes/bitrates is a plus
There's also opportunities for even better compression, since the
bitstream is still pretty redundant. On-air this isn't something you can
do much about since you want to keep latencies down. For storage I
believe just throwing gz or xz at codec2 data was enough to shrink it
considerably
/Tomas