Subject: Re: Re[2]: [linux-audio-dev] peakfiles and EDL's
From: Paul Kellett (paul.kellett_AT_maxim.abel.co.uk)
Date: Thu Mar 01 2001 - 20:59:34 EET
Robert Schrem <Robert.Schrem_AT_WiredMinds.de> wrote:
>
> > Speech waveforms in particular can be very asymmetric, so if you want
> > the peakfile display to look the same as the zoomed-out waveform you
> > need + and - peaks.
>
> Well, you will not be able to make percise zero crossin cuts
> from peak file information anyway. You would have to zoom in
> and then you will see the real sample data anyway.
I was talking about the asymmetry of speech waveforms over tens or
hundreds of cycles, not just from one half of a cycle to the next.
> The peak file 'waveform' display woun't look similar like the
> real data. For an explantation why, look for my detailed
> email a few day's back... :)
But the peak file *will* look like the real data...
Think of an oscilloscope - when the timebase is short you can clearly see
the waveform. As you turn the timebase to longer times, the waveform is
drawn very slowly across the screen, and you might not see each individual
cycle, just the envelope - the maximum and minimum ampitudes.
When you are zoomed out on a computer display, you don't want to waste
time drawing thousands of samples, up and down over the same column of
pixels, before moving to the next column and drawing thousands more - you
can save time and get the same result using a peakfile.
Paul.
-----------------------------
m a x i m | digital audio
http://www.maxim.abel.co.uk
paul.kellett_AT_maxim.abel.co.uk
This archive was generated by hypermail 2b28 : Thu Mar 01 2001 - 21:18:07 EET