Yesterday morning I
have uploaded a new version SlowVoice v0.2 to
I have reworked some details of the timing, and
added simplified scripts for receive-only and transmit-only in mono mode.
The serial port PTT control utility has also been modified
to function under WinXP.
This afternoon I am sending a few slow voice tests with
3 times deceleration in 1-minute frames. The 800 Hz wide
passband is centered on 137.0 kHz, thus 136.6 to 137.4
kHz corresponds to 0.3 to 2.7 kHz audio. Attached
screenshot shows the narrow SSB signal on the Twente WebSDR
tuned to 135.5 kHz USB. I used a narrow passband to
improve rejection for DCF and HGA. This audio around fc=1500
Hz was fed to the receive-only utility, which then
produced clear speech during 20 seconds of
each minute.
I intend to run some more test transmissions this afternoon,
starting now 14 UT. Any reports would be much
appriciated!
Best 73,
Markus (DF6NM)
-----
Original Message -----
Sent: Tuesday, January 08, 2013 11:31 PM
Subject: LF: Slow Voice utility
A few minutes ago I have uploaded an
experimental SlowVoice script to
The main purpose is to be able to
convey high quality speech in less than 800 Hz on the RF side, as
prescribed by German LF and MF regulations. By default, an audio
band of 300 to 2700 Hz is slowed by a factor of 3 and upconverted to 1500 Hz
center frequency. Thus a range of 1200 to 2000 Hz is fed to the
SSB transmitter. As an aside benefit we get a very much welcome 5
dB improvement in SNR.
Like JT9-1, the procedure is clock-synchronized
in fixed one-minute time slots. At the beginning of the minute, the
transmitting station is asked to speak for 20 seconds. It feels a bit
like to speaking to a telephone answering machine ;-) The slow
converted output of the growing audio buffer begins to play immediately, and
will go on during the whole minute. On the receive side, fast replay
begins at second number 40, and the shrinking buffer will finally be
used up by the end of the minute.
The SlowVoice.bat script is set up for full
duplex operation. One channel is supposed to handle the transmit side (eg.
mic connected to left input, transmitter to left output), while
the other channel has the receiver fed to the input and phones on
the output. This is symmetrical, and there is no further need to switch
audio pathways between TX and RX. It is also possible to monitor the
result of one's own transmission, eg. by physically connecting left
line out to right in. If you want you can use SpecLab in
conjunction with Virtual Audio Cable to preprocess both the mic
and receiver audio inputs. In this case VAC will have to be selected as the
standard audio input device before the running the script.
Whether a slots is used for TX
or RX is not controlled by the software but only by the user's PTT. I
have included two extra scripts which are supposed to activate the
serial handshake lines (RTS and DTR) on either even or odd
minutes. Unfortunately the port control does not seem to work on
my XP machine with an external USB to serial converter, so for the time
being you may have to do with either vox control or
manual switching.
I haven't gotten around writing more
detailed instructions, but some basic hints have been included in the
SlowVoice script. Similar to the first slow
WSPR experiments, the whole deceleration / acceleration process is based
on Wolf's sound utilities which you will need to download
from the DL4YHF web site. You will
have to run the included patch utility once to duplicate SndInput and
SndOutpt instances.
Have fun! And please report bugs and other
troubles.
73, Markus (DF6NM)
-----
Original Message -----
Sent: Sunday, January 06, 2013 1:50 PM
Subject: LF: Bandlimited SSB test on 136 kHz
...
Slow voice transmission (ie the audio
deceleration/acceleration method originally used by DK8KW and myself) would
be nicer as it can fit a full SSB channel into 800
Hz. I have been working on a semi-automatic control,
with a fixed one minute raster similar to JT9-1. This will hopefully
allow us to comfortably exchange one 20 second voice message per time slot
(speak during seconds 0 to 20, concurrent
transmit and receive at 1/3 speed from 0 to 60,
replay starting 40 to 60). Anyone interested?
...
