Jonathan,

>From the protocol:

"The error correction signal is interleaved with the packet data with a 
signaling rate of 1/2, a folding length of 3, and a depth of interleave of 24,"

You are correct and I stand corrected. It's been a while since I read it.

Only the header has interleaving and if a voice frame fails correction it is 
repeated for the next 20 ms until a new decode is attempted.

Every 21st data frame has a sync pattern for training.

Thanks for the peer review,

Craig




--- In [email protected], "Jonathan Naylor" <naylo...@...> wrote:
>
> Hi Craig
> 
> > This contrained the final (4800 bps/6.25 kHz) air interface to a 2400 bps 
> > AMBE perceptual coder with 2/3 convolutional FEC. This left about ~1kilo 
> > bit a second of data on the air interface for usage as data (as any user 
> > saw fit!).
> 
> You're a little wrong here. D-Star uses the AMBE codec in block FEC mode. The 
> FEC used is two blocks of Golay (24,12,8) with the remaining 24 bits 
> unprotected. I presume this latter block is of less significant bits that can 
> sustain corruption with relatively little impairment of the carried speech.
> 
> Jonathan  G4KLX
>


Reply via email to