[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [linrad] Feedback about the voicelab tests

Hi Pierre,

It is interesting that the results you find differ significantly
from what I have found myself. Did you use head-phones or were you
just listening off the loudspeakers?

I was at the library and found a book "Single-Sideband Systems and
Circuits" written by people at Collins  radio. I think their chapter
on speech processing has been the basis for "common knowledge" in 
the field and I do not think it is correct. The conclusions may
be reasonably ok, but the theoretical justification is not.

I will probably dig a bit deeper into this later on. In the meantime
I am interested if you or anyone else can find any advantage with
unusual processing. Things like using a large bandwidth for the 
microphone signal, pre-emphasis to amplify fast transcients,
clipping them, then de-emphasis and band pass to set a suitable 
frequency response. Such pre-processing of course followed by the
usual RF or AF clipping. The idea is the same as for noiseblanking.
Remove pulses in a high bandwidth rather than allow them to become
lengthened in a selective filter. 

The Linrad voice-lab allows such experiments and many others. I have
played a little myself with it and I did not notice any advantage
with anything. I would be interested to know if you can find any
parameter combination that is better than voicepar9. Maybe you find
7 or 8 better?

I have several other things to do before taking up the tx problems
again so there is no hurry. As things stand right now I think the
Linrad speech processing will be rather simple with a few parameters