[arm-allstar] ARM-allstar Digest, Vol 82, Issue 22

[David McGough]

Patrick,

The FIR low-pass filter that HamVoIP uses is a "brick wall" type filter;  
within practical computational limits, of course.  The filter corner
frequency is about 3500Hz. However, the initial attenuation slope is
gradual, being down by about 10dB at 3800Hz. As you get to 3940Hz the
downward slope is now nearly vertical. The ultimate attenuation is greater
than 68dB, reaching or exceeding that level of attenuation between 3990Hz
and 24KHz.

Voice grade CODECs readily take advantage of this extra "fidelity" and
sound great!


73, David KB4FXC


On Fri, 19 Mar 2021, "Patrick Perdue via ARM-allstar" wrote:

> Yeah, I'm aware of the Nyquist frequency and all that. Before I started 
> using Allstar, I would sometimes control nets on DMR with a DVStick and 
> my own processing, using a multi-band dynamics processor in what 
> basically amounted to a de-esser in reverse to dynamically increase the 
> energy in that range. People couldn't understand why I sounded so much 
> better than the typical DMR.
> 
> Relating this back to HamVoIP, I thought the LPF was a bit below 
> Nyquist, maybe around 3.5 kHz or so, which is why I was confused that 
> frequencies in the range of 3.8 were still passing.
> 
> Now that I've read more on that filter's implementation, it's less 
> surprising to me. I thought it was more of a cut-off, which isn't the case.
> 
> 
> On 3/19/2021 3:48 PM, "Ken via ARM-allstar" wrote:
> >>> I remember, some time ago, you mentioned that simpleusb has filters to
> > prevent anti-aliasing. Does this only apply to TX audio?
> >
> > The data on the receive side cannot include any audio components above 4
> > kHz, so no similar anti-alias filters are needed.  The USB fob D/A converter
> > is driven at 48 kHz, so the filtering you need is up at that frequency range
> > and hardware-based (after the D/A converter).
> >
> > On the transmit side I find it helpful to limit the energy starting at 3 kHz
> > because there are  just not very many bits/sec available for digitizing that
> > energy in that range and the result is in increasing distortion products
> > (not including aliasing).
> >
> > This is at odds with audio from YSF whose vocoder does some magic to pump
> > through a lot of energy in that audio frequency range .... (vocoder vs
> > codec).
> >
> > Ken/KE2N
> >
> > _______________________________________________
> >
> > ARM-allstar mailing list
> > ARM-allstar at hamvoip.org
> > http://lists.hamvoip.org/cgi-bin/mailman/listinfo/arm-allstar
> >
> > Visit the BBB and RPi2/3/4 web page - http://hamvoip.org
> _______________________________________________
> 
> ARM-allstar mailing list
> ARM-allstar at hamvoip.org
> http://lists.hamvoip.org/cgi-bin/mailman/listinfo/arm-allstar
> 
> Visit the BBB and RPi2/3/4 web page - http://hamvoip.org
>