Am 17.04.2013 22:05, schrieb David Rowe:
> Yes I think something like 800 bit/s is possible at roughly the same
> quality using Vector Quantisation of the LSPs.  However this wouldn't be
> robust to channel errors.
Well, I do intent to decrease bandwith in order to make it free for
error-correction (Viterbi FEC 3/4 and 2/3). The goal is to keep
''payload'' (Voice and eventiually data, too) under 1200 bit/s...

(It's part of the cryptofon-project I'm on... At first, it is intended
to work on 2400 bit/s, and carry vocoder, narrowband data transmissions
and of cource sync data to enshure stable end-to-end communication...,
including FEC in that bandwith! Codec 2 was the only solution that
seemed practicable w/o licensing a propietary/patented codec, and there
has already been an implementation on least-cost DSPs from TI's
Stellaris family.)

Vector Quatization seems pretty more predictable than an upper-level
compression of the transmission, which does not seem to be capable to
guarantee a fixed, low rate... But I'm not shure on that point, and may
have to take a look on compression standards like V.42bis and algorithms
such as Bzip2...

There needs to be testing on how far this can go. I heard the record is
300 bit/s for a vocoder - but at poor quality. (Guess that's the cost of
ultra-narrow-bandwith). But a 600 bit/s mode of Codec2  (or 800 bit/s as
first step) would be a huge breaktrough, finally killing any arguments
of propietary vocoders.


---Netzblockierer


>
> - DAvid
>
> On Wed, 2013-04-17 at 21:39 +0200, Netzblockierer wrote:
>> Hi,
>>
>> just asking: Besides, as Codec2 is a Multiband-Exitation-Codec, I'm
>> interested to know whether there can be even lower bandwiths achieved.
>> Of course, this may increase encoder and decoder complexity, but when
>> looking on ultra-narrowband - applilications, I only found a very few
>> competitors:
>>       * TWELP (seems patent pending)
>>       * MELPe (Thales has patent on the 600 bit/s version. all higher
>>         bandwiths are partly patented by other companies, like TI,
>>         Microsoft, etc. ... The Thales patent has been granted in 2005
>>         - so until 2030 it's out of reach...)
>>       * Two propietary encoders from DSP Innovators inc. - working
>>         with 600 and 800 bit/s
>>         http://dspini.com/index.php?com=page&sec=products, which are
>>         claimed patent-free and royality-free...
>>
>> I just read a while ago that there has been a testing on packet/frame
>> loss... It resulted in two typed of bits... One called ''Wo_E'' or
>> somelike that that are about 800 somewhat bit/s ... I suggested to try
>> what happens if all bits except these are ''lost''/ignored... Has
>> anyone tried it, and if, what are the results???
>>
>>
>> Greetings,
>> Netzblockierer
>>


------------------------------------------------------------------------------
Precog is a next-generation analytics platform capable of advanced
analytics on semi-structured data. The platform includes APIs for building
apps and a phenomenal toolset for data science. Developers can use
our toolset for easy data analysis & visualization. Get a free account!
http://www2.precog.com/precogplatform/slashdotnewsletter
_______________________________________________
Freetel-codec2 mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Reply via email to