From: Rik Strobbe <rik.strobbe@kuleuven.be>
To: "rsgb_lf_group@blacksheep.org" <rsgb_lf_group@blacksheep.org>,
        "rsgb-lf-group@groups.io" <rsgb-lf-group@groups.io>,
        "rsgb_lf_group@yahoogroups.co.uk" <rsgb_lf_group@yahoogroups.co.uk>,
        "k3mf@aol.com" <k3mf@aol.com>
Thread-Topic: JT9-5 struggle
Thread-Index: AQHUn5Ebn6R6fzr6d0S1bCBzP3P+zQ==
Date: Sat, 29 Dec 2018 16:37:16 +0000
Message-ID: <1546101436080.48398@kuleuven.be>
References: <565118912.10306055.1546030653480.ref@mail.yahoo.com>,<565118912.10306055.1546030653480@mail.yahoo.com>,<1546076436128.54396@kuleuven.be>
In-Reply-To: <1546076436128.54396@kuleuven.be>
Accept-Language: nl-BE, en-GB, en-US
Content-Language: nl-BE
MIME-Version: 1.0
Subject: LF: JT9-5 struggle
Content-Type: multipart/alternative;
	boundary="_000_154610143608048398kuleuvenbe_"
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org

--_000_154610143608048398kuleuvenbe_
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

As mentioned earlier I have been struggling to get JT9-5 working for over a=
 week.

Here is why:

With JT9-2 I noticed that it decoded fine but always gave a DT value of -0.=
5 or -0.6

This was inherent to the resampling used: As any JT9 transmission JT9-2 sta=
rts at 1 seconds after the cycle start, so after resampling at a rate of 20=
/9 (=3D 2.222) the transmission will start at 0.45 seconds (1/2.222). This =
was easily fixed by inserting 6600 "zero's" (=3D 0.55 seconds at 12kHz samp=
le rate) at the beginning of the audio file sent to the decoder.

So for JT9-5 I did the same, now the resampling rate was 160/27 (=3D 5.923)=
 and after resampling the transmission would start at 0.16875 seconds (27/1=
60), so now 9975 zero's were inserted to fix the timing error.

But JT95-5 did not work properly: except for an odd case no decodes (but wh=
en it decoded it decoded well, even at very low SNR).

After one day of struggling I decided tot make a small app that

- created a JT9-5 signal at any given SNR

- resampled to JT9-5 signal

- decoded the resampled audio file

In the app everything worked fine instantly and even after 10 tems double c=
hecking for differences bewteen the code in this app and SlowJT9 I couldn't=
 find any ... until at last I notoced that I didn't implement the timing co=
rrection in the test app.

After removing the timing correction in SlowJT9 JT9-5 worked fine.

I did some further testing and noticed that with a timing correction up to =
about 0.6 seconds (7200 samples) everything works fine, but from then on de=
grading starts and very rapidly nothing is decoded anymore.

So for now the timing correcting is disabled. Apart from the fact thta for =
most transmissions negative DT values will show up this does not affect the=
 functionality of SlowJT9.

But I am still puzzled why this happens?

Any idea's?


73, Rik  ON7YD - OR7T


--_000_154610143608048398kuleuvenbe_
Content-Type: text/html; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<html>
<head>
<meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3Diso-8859-=
1">
<style type=3D"text/css" style=3D"display:none"><!-- p { margin-top: 0px; m=
argin-bottom: 0px; }--></style>
</head>
<body dir=3D"ltr" style=3D"font-size:10pt;color:#000000;background-color:#F=
FFFFF;font-family:Tahoma,Geneva,sans-serif;">
<p>As mentioned earlier I have been struggling to get JT9-5 working for ove=
r a week.<br>
</p>
<p>Here is why:<br>
</p>
<p>With JT9-2 I noticed that it decoded fine but always gave a DT value of =
-0.5 or -0.6&nbsp;</p>
<p>This was&nbsp;inherent to the resampling used: As any JT9 transmission J=
T9-2 starts at 1 seconds after the cycle start, so after resampling at a ra=
te of 20/9 (=3D 2.222) the transmission will start at 0.45 seconds (1/2.222=
). This was easily fixed by inserting 6600
 &quot;zero's&quot; (=3D 0.55 seconds at 12kHz sample rate)&nbsp;at the beg=
inning of the audio file sent to the decoder.<br>
</p>
<p>So for JT9-5 I did the same, now the resampling rate was 160/27 (=3D&nbs=
p;5.923) and after resampling the transmission would start at 0.16875&nbsp;=
seconds (27/160), so now 9975 zero's were inserted to fix the timing error.=
<br>
</p>
<p>But JT95-5 did not work properly: except for an odd case no decodes (but=
 when it decoded it decoded well, even at very low SNR).<br>
</p>
<p>After one day of struggling I decided tot make a small app that&nbsp;</p=
>
<p>-&nbsp;created a JT9-5 signal at any given SNR<br>
</p>
<p>- resampled to JT9-5 signal<br>
</p>
<p>- decoded the resampled audio file<br>
</p>
<p>In the app everything worked fine instantly and even after 10 tems doubl=
e checking for differences bewteen the code in this app and SlowJT9 I could=
n't find any ... until at last I notoced that I didn't implement the timing=
 correction in the test app.<br>
</p>
<p>After removing the timing correction in SlowJT9 JT9-5 worked fine.<br>
</p>
<p>I did some further testing and noticed that with a timing correction up =
to about 0.6 seconds (7200 samples) everything works fine, but from then on=
 degrading starts and very rapidly nothing is decoded anymore.<br>
</p>
<p>So for now the timing correcting is disabled. Apart from the fact thta f=
or most transmissions negative DT values will show up this does not affect =
the functionality of SlowJT9.<br>
</p>
<p>But I am still puzzled why this happens?<br>
</p>
<p>Any&nbsp;idea's?<br>
</p>
<p><br>
</p>
<p>73, Rik&nbsp; ON7YD - OR7T<br>
</p>
<p><br>
</p>
</body>
</html>

--_000_154610143608048398kuleuvenbe_--