----- Original Message -----
From: Ian White, G3SEK <G3SEK@ifwtech.demon.co.uk>
To: <amps@contesting.com>
Sent: Wednesday, March 29, 2000 9:21 PM
Subject: [AMPS] Re: Serious debate: Dynamic IMD testing
>
> Petter wrote:
> >
> >Observe the change in topic...
> >
> In fact, there is a serious point behind my suggestion.
>
> >
> >"Ian White, G3SEK" skrev:
> >> The world clearly needs a standard reference .wav file.
> >>
> >> Now who shall we nominate?
> >
> >Shall we take someone from Japan , deep Texas or even Wales ??
> >
> >No I think we need to define a dynamic test method that simulates the
human
> >voice.
>
> Devising a test signal that adequately simulates the characteristics of
> the human voice is extremely difficult - and even when you've done it,
> it still suffers from the same problem of "whose voice are you
> simulating?"
>
> I think there could be two types of test - a simple tone-based one, and
> a totally realistic test based on real voices. I just don't see much use
> for the in-between cases using elaborately synthesized test signals,
> because in the end it's performance with real voices that interests us.
>
> 1. I think we all agree that the classic two-tone test is too simplistic
> because it doesn't exercise the power supply regulation in the way a
> real voice does. So let's have a two-tone signal, with envelope
> modulation of both tones at a swept frequency from say 1Hz up to several
> hundred Hz. If you keep the separation of the first two tones constant,
> their IMD products would stay in the same place on the spectrum analyser
> screen. This allows you to see how each IMD product level varies with
> modulation frequency.
>
> 2. But why bother with building up a more complex simulation of a human
> voice when we can have the real thing - or indeed several human voices?
> The idea for this test would be to set the spectrum analyser to peak-
> hold mode and scan for a long time to build up a picture of the highest
> IMD levels there have been at any frequency. If you choose the right
> recorded material, you can use as many different voices as you have time
> to run the test for, and keep building up the IMD picture.
>
>
> The signal source for Test 1 could be built in hardware, but this seems
> a waste of time when we have PCs with sound cards. Does anybody know how
> to program a sound card to generate Test 1 - or even a basic two-tone
> test with one tone out of each L/R channel?
>
I think that I have a program that will allow me to do this.Tell me exactly
what is needed.I think 2 fixed interval tones and a swept one can be done
with software designed to emulate old moog synthesisers.I mean the musical
type.
These programs allow an unlimited number of oscillators to be "patched" and
each oscillator can be modulated both with respect to frequency and
amplitude.It is not normal to use sine waves which have no harmonic
relationship for musical purposes but the program designers know we all want
to make Sci Fi movie noises anyway so have left it open ended.
Let me know the frequecies of the 2 fixed tones and the sweep range of the
third tone and I will create the wav file.
If you get a wav file of someone speaking I can pitch shift it and change
the formants to make different harmonic empahasis without altering the
length of the file.This would simulate the natural variations in peoples
voices quite well.
As for us agreeing on what the standard should be, that is a different
matter ....
73 de Conrad G0RUZ
--
FAQ on WWW: http://www.contesting.com/ampsfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|