|
Hi Stefan,
I don't really know the answer but here's a guess:
If I remember right, your MF receiver
gets the signal from the antenna through a mechanical filter. The
passband is some 8 or 10 kHz wide, occupying only a part of the whole 24
kHz Nyquist range. Outside the passband there will still be some noise from
the RX (and possibly also weak NDB signals) but at much lower level. I suspect
that if you turn up the volume, a few of these weak channels may rise above
a threshold where the Vorbis encoder decides to retain the information. If
this is correct, the increased bitrate would mainly be from out-of-band
information, and a lower level would not compromise the in-band
performance.
One way to check the hypothesis is to
view the full band in a relatively fast spectrogram, with low contrast to show a
large dynamic range. Attached is a
spectrogram from the lovely song "Dreams" by the Cranberries,
replayed from a 128 kbit/s mp3 file. The output from Mediaplayer was routed
to Spectrogram software through the analog mixer. You can see
many black patches, which correspond to temporary low levels in some audio
channels. These have obviously been cut out by the encoder to reduce
bitrate. It's interesting to note that although the nominal samplerate
of this mp3 is 44.1 kHz, most of the high frequency content above 16 kHz has
also been dropped, except for occasional louder
segments.
Best 73,
Markus (DF6NM)
Sent: Saturday, June 20, 2015 9:04 PM
Subject: LF: VLF vorbis stream, Question
Hi
all,
There are vorbis streams available on VLF by Paul Nicholson and
Wolf/DL4YHF and a few more known people from the scene. As far as i know these
are in vorbis format, like my MF/VLF stream.
What i found yesterday, the data
rate depends slightly on the volume level.
Now i'm asking myselfe if
there is a 'quality' loss when using a low mic volume level ??
To become
more precise: The usual method is to check where the noise level of the
soundcard can be found, e.g. in Speclab it shows -120 dB for example. Then, if
one connects the RX, it may rise to -118 dB (in what ever FFT bin width and Mic
gain level). Then, connecting the antenna to the RX may let the (daytime!) noise
level rise to -100 dB. Then one knows that the daytime band noise level is 18 dB
above the noise level of soundcard+RX and everything is fine and
the dynamic range is somewhere near 100 dB or maybe just 90 dB but at least it
is high enough...
Can this method or thinking be applied when SpecLab is
getting its data via a vorbis stream???
I can detect the noise level
without an antenna connected and prove that it is about 20 dB lower as when the
antenna is connected. So i assumed that everything is all right. But when
playing with the mic gain level, i can see that the data rate rises about 10%
when adding another 20 dB. So is there a loss of data, resulting in a lower SNR
of incoming signals when using a low mic level, although it is still well enough
above the soundcard+RX noise?? (Of course i want to keep the mic level as low as
possible without a quality loss, to have a dynamic range as high as
possible)
I just noticed that effect last night and now i'm aksing if there
are unwanted losses.
73, Stefan
|
dreams_mp3_128kbs_short.jpg
Description: JPEG image
|
