[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
>From tothl Wed Jun 23 19:46:39 +0200 1999 remote from inf.u-szeged.hu
Date: Wed, 23 Jun 1999 19:46:39 +0200 (MET DST)
From: Toth Laszlo <email@example.com>
Subject: Re: Pascal per Hertz
Received: from inf.u-szeged.hu by inf.u-szeged.hu; Wed, 23 Jun 1999 19:46 MET
Content-Type: TEXT/PLAIN; charset=US-ASCII
On Wed, 23 Jun 1999, Paul Boersma wrote:
> Praat is not Matlab. The sound is in a Sound object, which has information
> about the time domain and about the sampling, and is considered to carry
> samples that represent air pressures in Pascal. The "fft" converts this
> Sound object into a Spectrum object
FFT is not Matlab. FFT is not Praat. FFT (or, rather, DFT) is a formula that
converts a series of numbers into another series of numbers. For me.
That's why I didn't get your previous mail. Sorry. No I see that by
"Praat fft" you meant "the spectrum analysis facilities (and the belonging
objects) in the Praat software package".
(At first I thought "Praat FFT" was something like "Bruun FFT" or
> In Praat, the maximum absolute value of the samples is 1, which, if considered
> Pascals, corresponds to an intensity of 94 dB re auditory threshold.
> [...] you should use a dB meter once,
> with your computer set to a maximum input gain, and use the same gain
> setting for all subsequent recordings.
Sounds uncomfortable to give a dB meter with each copies of a speech
recognizer (well, usually you'll have to buy a special mike with it
anyway, so...). Or should we avoid non-linear preprocessing?
Laszlo Toth * "I don't like people
Hungarian Academy of Sciences * working all day,
Research Group on Artificial Intelligence * only working just to
e-mail: firstname.lastname@example.org * see next tomorrow"
http://www.inf.u-szeged.hu/~tothl * /Accept/