Subject: Re: [ecasound] reverbs
From: Jeremy Hall (jhall@uu.net )
Date: Sat Sep 11 1999 - 18:52:02 EEST
continuing the audit:
-efh and -efl: they may not be doing the right thing
-efh:1000 -efl:1000 does not produce silence. It should, or in the least
should produce an extremely narrow frequency set on 1000hz. -efh:1000
-efl:1100 in my mind should definitely produce silence because in the
first pass, you say crop all signal that is below 1000hz, and in the
second filter set you say crop all signal above 1100hz, clearly getting
that small overlap signal. Adding some band reject filters at 1100 and
1000 doesn't help.
-efi: appears to do something interesting if one plays with the numbers,
but the effect is not quantifyable, meaning, it is difficult to predict
what setting the value is going to do, mainly because I do not fully
understand the effect.
-efl:cutoff_freq
Lowpass filter. Only frequencies below 'cut-
off_freq' are passed through.
ah it seems my overlap was going the wrong way. :(
ok, trying -efh:1000 -efl:1100
but notice the incongruency
ecasound -c -i /tmp/in.raw -efhl:1100 -efh:1000
****************************************************************************
* ecasound v1.5.9r5 (C) 1997-1999 Kai Vehmanen
*
****************************************************************************
(eca-session) Interactive mode enabled.
- [ Chainsetup/Adding a new chain operator ]
-------------------------------
FX: Highpass filter enabled, cutoff frequency is 1100.00.
- [ Chainsetup/Adding a new chain operator ]
-------------------------------
FX: Highpass filter enabled, cutoff frequency is 1000.00.
- [ Chainsetup/Adding a new input ]
----------------------------------------
(eca-chainsetup) Assigning file to chains: default
(eca-chainsetup) Opening file "/tmp/in.raw" for reading.
(audio-io) Channels 2, srate 44100, bits 16, align 4, bytes second 176400.
- [ Chainsetup/Adding a new output ]
---------------------------------------
(eca-chainsetup) Assigning file to chains: all
(audio-io) Opening file "/tmp/out.raw" for reading and writing.
(audio-io) Channels 2, srate 44100, bits 16, align 4, bytes second 176400.
(eca-session) Setting chainsetup "command-line-setup" active.
ecasound ('h' for help)> - [ Engine/Initializing ]
--------------------------------------------------
- [ Engine/Mixmode "simple iactive" selected ]
-----------------------------
t
- [ Controller/Processing started ]
----------------------------------------
WARNING packet 0 out of order. old value 11925 (diff -11925)
ecasound ('h' for help)>
this produces some signal that is consentrated around 1000hz.
anyway, the point is that something isn't appearing to do what I think it
should do. Maybe I am misunderstanding audio mechanics in a big way,
because I just tried -efs:15000,5000 and got all sorts of signal that did
not sound to be in this range.
-efr:
My current theory is that -efr:1000,500 will reject 1000 to 1500, but not
1000 to 500. I need to test this theory a little more.
-gc: works fine
-ge: I could not get this to work, it complained of argument problems.
-kos: OH WIERD!
but warnings are initiated by the example #1.
ecasound -i /tmp/in.raw -ef3:800,1.5,0.9 -c -kos:1,400,4200,0.2,0
-kos:2,0.1,1.5,0.15,0
****************************************************************************
* ecasound v1.5.9r5 (C) 1997-1999 Kai Vehmanen
*
****************************************************************************
(eca-session) Interactive mode enabled.
- [ Chainsetup/Adding a new chain operator ]
-------------------------------
FX: Resonant lowpass filter enabled, cutoff frequency is 800.00,
resonance 1.50 and overall gain of 0.00.
- [ Chainsetup/Adding controller source ]
----------------------------------
(osc-sine) Sine oscillator created; frequency 0.2 and initial phase of
0.0.
WARNING: Chainop type "Resonant Lowpass filter" doesn't yet support
dynamic parameter controlling.
(gcontroller) Created a new general controller; chainop "Resonant Lowpass
filter" source "Sine oscillator" param number 1 range from 400.00 to
4200.00.
- [ Chainsetup/Adding controller source ]
----------------------------------
(osc-sine) Sine oscillator created; frequency 0.1 and initial phase of
0.0.
WARNING: Chainop type "Resonant Lowpass filter" doesn't yet support
dynamic parameter controlling.
(gcontroller) Created a new general controller; chainop "Resonant Lowpass
filter" source "Sine oscillator" param number 2 range from 0.10 to 1.50.
- [ Chainsetup/Adding a new input ]
----------------------------------------
(eca-chainsetup) Assigning file to chains: default
(eca-chainsetup) Opening file "/tmp/in.raw" for reading.
(audio-io) Channels 2, srate 44100, bits 16, align 4, bytes second 176400.
- [ Chainsetup/Adding a new output ]
---------------------------------------
(eca-chainsetup) Assigning file to chains: all
(audio-io) Opening file "/tmp/out.raw" for reading and writing.
(audio-io) Channels 2, srate 44100, bits 16, align 4, bytes second 176400.
(eca-session) Setting chainsetup "command-line-setup" active.
ecasound ('h' for help)> - [ Engine/Initializing ]
--------------------------------------------------
- [ Engine/Mixmode "simple iactive" selected ]
-----------------------------
t
- [ Controller/Processing started ]
----------------------------------------
WARNING packet 0 out of order. old value 9509 (diff -9509)
ecasound ('h' for help)>
Clearly this wants more explanation so the full potential is realized. (I
realize documentation isn't everybody's favourite task)
_J
Kai Vehmanen said:
>
> On Wed, 8 Sep 1999, Jeremy Hall wrote:
>
> > Does the band filter (efb) work correctly?
>
> Well, it's not perfect, but does the job most of the time.
>
> > Some of the other band filters produce squealing noises. You'd think that
> > the combination
> [...]
> > -efs:440,6000 -efr:440,6000
> > and
> > -efb:440,6000 -efr:440,6000
>
> It seems that the band-reject filter "-efr" isn't working.
> On the other hand, the above combintion might still not work,
> as you've used a freq-range of -2560Hz->3440Hz (440Hz being
> the center-freq).
>
> Btw; to everybody, if you think some effect is behaving weirdly,
> notify me about it! I've had to do some big changes to the effect
> class-design and haven't yet had the time to test all effects.
> Usually these problems are caused by uninitialized class members
> or something similar. I rarely have to touch the actual
> effect algorithms so these are easy to fix.
>
> > would produce silence. They don't. I seem to misunderstand the cutoff
> > filters, because I seem to hear pitches in the red zone but distant--am I
> > hearing harmonics in the green zone?
>
> Well, simple filters like these don't have a very sharp cutoff.
> What parameteres did you use?
>
> --
> Kai Vehmanen <kaiv@wakkanet.fi> -------- CS, University of Turku, Finland
> . http://www.wakkanet.fi/ecasound/ - linux audio processing
> . http://www.wakkanet.fi/sculpscape/ - ambient-idm-rock-... mp3/ra/wav
>
>
>
>
>
This archive was generated by hypermail 2a24 : Sat Sep 25 1999 - 19:41:11 EEST