----- Original Message -----
Sent: Thursday, January 03, 2013 12:33
PM
Subject: Re: LF: SSB - why not go
digital?
Hi Geri, all,
I like the idea
of trying voice on MF or even LF, and I was fascinated by actually
hearing Gus talking on 507 kHz, and Geri on LF slow
voice. I admit that I even tried (low-power daytime) SSB above
135.7 kHz once, and Stefan was actually able to hear and record my voice,
albeit at marginal SNR at 180 km. As my LF TX antenna is fairly narrowband
(Q ~ 200), I had inserted a phase shifter to transform the low output
impedance of the PA into a current source at the feedpoint.
The good thing
about slow voice is that we also get the SNR gain associated with the narrower
bandwidth. But the non-realtime operating procedure is sort
of difficult. We did use "roger beeps" to mark the end of
message, but there is always an inconvenient delay for recording and
replaying the messages. So some sort of realtime narrowband voice
transmission would indeed be desirable.
However I would
NOT fancy digital voice modes at all. For one, like all digital
modes it's all-or-nothing. Either the message decodes
well, or you get garbage, and you don't know what has been going on in
the channel. In linear analog SSB, you hear if someone else is calling on
frequency, or what type of interference came up, or whether there is
selective fading. The other thing I don't like about digital modes is
that they tend to occupy the whole channel permanently with high average power
- just look at the spectra of DRM vs AM modulated BC transmissions. It
may well turn out that a 2 kHz linear SSB transmission is much more
friendly than a 1 kHz digital channel.
I have been
making some attempts on analog quasi-realtime narrowband voice transmissions.
The principle was is much the same that has long been used
for changing audio speed without changing pitch, ie cut the audio
into ~ 20 ms grains. To accellerate or reduce bandwidth, you
either leave out or average several grains, using a sliding window.
On replay, each grain is repeated to extend it in time. This is not
difficult to implement. In my own trials, I could maintain speech readability
with a factor four bandwidth reduction, but it did sound very "robotic"
because the fixed timing for the grains impressed itself onto the
voice pitch. I think that there are better ways of adapting
the timing by tracking the fundamental frequency to preserve
the pitch modulation. Using Windows media player for
accelerating and decelerating seemed to provide quite natural
speech quality at a reduction factor of four.
Regarding the
frequency allocations, like Graham I would think that on MF the narrowband
beacons should best be placed in narrow slots near the band edges. To avoid
blocking, we would preferably again use split band for west-east vs
east-west intercontinental work, as we do on LF. Parts of these slots should
also be utilized for the slow versions of Opera or WSPR. The wider modes
would better be placed in the in the middle, perhaps with 2 kHz
each for CW, middle-range digital modes, and one
full-width SSB channel.
Best 73,
Markus
(DF6NM)
...
