El 24/02/19 a las 23:15, Gregory Maxwell escribió:
> Are there really many places where spectral efficiency is actually
> important for VHF+ ham communications? It was my impression that ham
> DV systems were pushing on spectral efficiency only because of being
> receiving commercial system
On Sun, Feb 24, 2019 at 9:24 PM Jeroen Vreeken wrote:
>
> Hi David,
>
> 52bits per 30ms might work out nicely.
>
> 30ms at 2400baud is 72 bits. If we use a 16 bit uw there are still 4
> bits to spare.
> I don't think FEC is that important when using an existing FM system.
> For more advanced work
Hi David,
52bits per 30ms might work out nicely.
30ms at 2400baud is 72 bits. If we use a 16 bit uw there are still 4
bits to spare.
I don't think FEC is that important when using an existing FM system.
For more advanced work with TDMA that will help, but I think that would
require its own
Hello Jeroen,
The experimental quantisation I'm working on is flexible, you can trade
off the number of VQ stages and frame rate (10,20,30 ... ms).
In order to get a FreeDV mode on the air, I've settled on 52 bits every
30ms (1733 bits/s). You'll need syncronisation and maybe FEC on top of
Hi Steve - yes that's including FEC parity bits, and modem
sycnronisation overhead like pilot symbols. The quantised LPCnet codec
in that example is 1733 bits/s.
- David
25/02/19 04:47, Steve wrote:
> Jeroen,
>
> I noticed that David updated his modem design file:
>
>
Jeroen,
I noticed that David updated his modem design file:
https://sourceforge.net/p/freetel/code/HEAD/tree/codec2-dev/octave/modem_codec_frame_design.ods?format=raw
Lot's of parameters and stuff in there.
Using the ofdm_mod and octave ofdm_rx with those parameters, I get as
output:
Rs=48.78
Hi David,
How 'fixed' is the 'around 2000 bits/s' number? And do you have some
idea of the frame size you are going to use?
(e.g. 40ms or something different?P
I really liked the sound of the new modes and would like to test them on
UHF with a mode based on 2400B.
If we take the current 2400B
Hi David,
If the LPCNet codec can run without additional dependencies, that'd be
great! We have the opportunity to train it for specific speakers when we
scrape content from educational sites on our server. The only issue is
the size of the model.
Unfortunately we don't have much expertise on c
Hi Mike,
Unfortunately the masking model work didn't lead to a viable Codec 2 mode.
I think a LPCNet codec might suit your application well, I'll have a
release in the next few weeks for you to play with (around 2000 bits/s).
Jean-Marc Valin (the author) has ported the code to C and to run on
Hi Codec2 list,
I'm working together with Samih, looking at shrinking Khan Academy and
other educational content for our offline library app. I've been trying
to figure out optimal codec2 encoding / decoding parameters.
We know who the speaker is in each clip. As far as I can understand, the
10 matches
Mail list logo
