Archived from groups: rec.audio.tech (More info?)

I have a WAV file of spoken voice, punctuated by applause. The
audience is very polite, they don't applaud while the person is
talking. However, the mike picked up the applause very loudly. While
the levels of the spoken voice are all pretty low (varying from person
to person), the applause and sometimes laughter is right out to the
limits.

What I want to do is to take all sounds above a certain threshold
(which would be the loudest that the speaking voices get) and cut its
volume by an amount that I determine. Say 18 dB. So in effect, I'd be
deamplifying only laughter and applause.

I know that I can get so-so results using (Dynamic Range) Compression
(in Audacity) with Gain not applied afterwards. I also know that I can
do this manually, by selecting each instance of applause and processing
it by hand. But this is the computer age, so why can't I do it all at
one go? Another way to look at this concept is a kind of gating, with
the reduction being applied only to sounds which are ABOVE a certain
level rather than below. Audacity doesn't seem to have gating anyway,
or at least not under that name.

Does the concept I have outlined have a name?

The Compression filter in Audacity is numerated in dB, and the waveform
is numerated on a scale from -1 to +1. Is there a rule-of-thumb
equivalency between those two scales?

Thanks.

--
Jonathan Berry

Archived from groups: rec.audio.tech (More info?)

<jberry@islandnet.com> wrote in message
news:1127862813.783673.202630@g43g2000cwa.googlegroups.com...
> I have a WAV file of spoken voice, punctuated by applause. The
> audience is very polite, they don't applaud while the person is
> talking. However, the mike picked up the applause very loudly. While
> the levels of the spoken voice are all pretty low (varying from person
> to person), the applause and sometimes laughter is right out to the
> limits.
>
> What I want to do is to take all sounds above a certain threshold
> (which would be the loudest that the speaking voices get) and cut its
> volume by an amount that I determine. Say 18 dB. So in effect, I'd be
> deamplifying only laughter and applause.
>
> I know that I can get so-so results using (Dynamic Range) Compression
> (in Audacity) with Gain not applied afterwards. I also know that I can
> do this manually, by selecting each instance of applause and processing
> it by hand. But this is the computer age, so why can't I do it all at
> one go? Another way to look at this concept is a kind of gating, with
> the reduction being applied only to sounds which are ABOVE a certain
> level rather than below. Audacity doesn't seem to have gating anyway,
> or at least not under that name.
>
> Does the concept I have outlined have a name?
>
> The Compression filter in Audacity is numerated in dB, and the waveform
> is numerated on a scale from -1 to +1. Is there a rule-of-thumb
> equivalency between those two scales?


I have no knowledge of Audacity, but you are attempting to do exactly the
right thing, it is called compression or dynamic processing, and lots of
software
will do it very successfully. Try Soundforge, Cubase etc these have the
added advantage
of being able to use third party plugins if you are not happy with the in
built version

Dave H

Archived from groups: rec.audio.tech (More info?)

As I understand it, dynamic range compression sets a threshold, and
then volumes above that threshold are brought close to it, possibly
very close to it. The higher the signal ratio "Ratio", the closer the
output sound will be to the threshold, but never below it. Is that a
correct understanding?

What I was proposing is that volumes above the threshold be reducible
by a user-specified amount. In this case, I want to reduce them so
that they are BELOW the threshold. Is there such a thing? If so, what
is it called? If it is a form of dynamic range compression, then does
anybody know a compession filter which offers it?

I know that I can do this "by hand", highlighting the fifty or so
instances of pure applause in the clip, and processing them separately.
I'd rather have the computer do it, because the criterion is simple
and consistent.

Thanks.

--
Jonathan Berry

Archived from groups: rec.audio.tech (More info?)

A graphic compressor should allow you to draw such a curve.
Simply rising at 45-degree angle until you get to the volume
you want, then dipping to a level below that. I've been using
CoolEdit and NTrack, and I believe it's possible in both.
(although I haven't tried it myself).

Scott

jberry@islandnet.com wrote:
: As I understand it, dynamic range compression sets a threshold, and
: then volumes above that threshold are brought close to it, possibly
: very close to it. The higher the signal ratio "Ratio", the closer the
: output sound will be to the threshold, but never below it. Is that a
: correct understanding?

: What I was proposing is that volumes above the threshold be reducible
: by a user-specified amount. In this case, I want to reduce them so
: that they are BELOW the threshold. Is there such a thing? If so, what
: is it called? If it is a form of dynamic range compression, then does
: anybody know a compession filter which offers it?

: I know that I can do this "by hand", highlighting the fifty or so
: instances of pure applause in the clip, and processing them separately.
: I'd rather have the computer do it, because the criterion is simple
: and consistent.

: Thanks.

: --
: Jonathan Berry