Parametric FIR audio equalizer

Roman Rumian <usun_torumian@agh.edu.pl> wrote in message news:<ckm8dl$qvd$1@galaxy.uci.agh.edu.pl>...
> Hello,
> 
> maybe my question is stupid, but does such a device/algorithm exist in 
> practice ?
> 
> Regards
> 
> Roman Rumian

Depends upon what you actually mean by parametric in the context of an
FIR but this product exists ( HarBal Harmonic Balancer
http://www.har-bal.com ) and works well (IMHO), though I may be biased
since I wrote it. On the other hand many other people seem to have
found it useful.

Regards,


Paavo Jumppanen.

if you have Matlab , check the "invfreqz" function, it can do the job :

>> help invfreqz

 INVFREQZ  Discrete filter least squares fit to frequency response data.
    [B,A] = INVFREQZ(H,W,NB,NA) gives real numerator and denominator 
    coefficients B and A of orders NB and NA respectively, where
    H is the desired complex frequency response of the system at frequency
    points W, and W contains the normalized frequency values within the 
    interval [0, Pi] (W is in units of radians/sample).


however, i am not enough experienced to explain how it works .....

On 14 Oct 2004 16:09:57 -0700, PaavoJumppanen@iname.com (Paavo
Jumppanen) wrote:

>Roman Rumian <usun_torumian@agh.edu.pl> wrote in message news:<ckm8dl$qvd$1@galaxy.uci.agh.edu.pl>...
>> Hello,
>> 
>> maybe my question is stupid, but does such a device/algorithm exist in 
>> practice ?
>> 
>> Regards
>> 
>> Roman Rumian
>
>Depends upon what you actually mean by parametric in the context of an
>FIR but this product exists ( HarBal Harmonic Balancer
>http://www.har-bal.com ) and works well (IMHO), though I may be biased
>since I wrote it. On the other hand many other people seem to have
>found it useful.
>
>Regards,
>Paavo Jumppanen.

Hi Paavo,

  congratulations on your Har-Bal product.
I'm not an "audio" guy, but your Harmonic Balancer 
looks "neat!"

[-Rick-]
 

Hi Jerry,

U�ytkownik Jerry Avins napisa�:
(...)
>>U�ytkownik Jerry Avins napisa�:
(...)
> Thank you, Roman. I hope "u�ytkownik" isn't too derogatory. :-p
> 
> Jerry

sure - are you a "User" of anything ? :-)

Roman

U�ytkownik Paavo Jumppanen napisa�:
(...)
> Depends upon what you actually mean by parametric in the context of an
> FIR but this product exists ( HarBal Harmonic Balancer
> http://www.har-bal.com ) and works well (IMHO), though I may be biased
> since I wrote it. On the other hand many other people seem to have
> found it useful.

congratulation Paavo, but do you know this page:
http://www.octiv.com/index.asp?content=demo_real_music

How compare HarBal to Octiv Spectral Rebalancing ?

Could you send me a demo wav file ?

Regards

Roman

Roman Rumian wrote:

> Hi Jerry,
> 
> Użytkownik Jerry Avins napisał:
> (...)
> 
>>> Użytkownik Jerry Avins napisał:
> 
> (...)
> 
>> Thank you, Roman. I hope "użytkownik" isn't too derogatory. :-p
>>
>> Jerry
> 
> 
> sure - are you a "User" of anything ? :-)
> 
> Roman

Not of anything illegal! Thanks for the translation.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

r.lyons@_BOGUS_ieee.org (Rick Lyons) wrote in message news:<416fb506.385240421@news.sf.sbcglobal.net>...
> On 14 Oct 2004 16:09:57 -0700, PaavoJumppanen@iname.com (Paavo
> Jumppanen) wrote:
> 
> >Roman Rumian <usun_torumian@agh.edu.pl> wrote in message news:<ckm8dl$qvd$1@galaxy.uci.agh.edu.pl>...
> >> Hello,
> >> 
> >> maybe my question is stupid, but does such a device/algorithm exist in 
> >> practice ?
> >> 
> >> Regards
> >> 
> >> Roman Rumian
> >
> >Depends upon what you actually mean by parametric in the context of an
> >FIR but this product exists ( HarBal Harmonic Balancer
> >http://www.har-bal.com ) and works well (IMHO), though I may be biased
> >since I wrote it. On the other hand many other people seem to have
> >found it useful.
> >
> >Regards,
> >Paavo Jumppanen.
> 
> Hi Paavo,
> 
>   congratulations on your Har-Bal product.
> I'm not an "audio" guy, but your Harmonic Balancer 
> looks "neat!"
> 
> [-Rick-]

Thanks Rick,

We had a bit of an infamous start about a year ago being branded snake
oil merchants. The dust has settled and it seems that people can now
see merit in the approach. It's a start, but there's plenty more left
to do and I'm, snowed under with work. Oh well.

Paavo.

Roman Rumian <usun_torumian@agh.edu.pl> wrote in message news:<ckp8tq$d6s$1@galaxy.uci.agh.edu.pl>...
> U&#4294967295;ytkownik Paavo Jumppanen napisa&#4294967295;:
> (...)
> > Depends upon what you actually mean by parametric in the context of an
> > FIR but this product exists ( HarBal Harmonic Balancer
> > http://www.har-bal.com ) and works well (IMHO), though I may be biased
> > since I wrote it. On the other hand many other people seem to have
> > found it useful.
> 
> congratulation Paavo, but do you know this page:
> http://www.octiv.com/index.asp?content=demo_real_music
> 
> How compare HarBal to Octiv Spectral Rebalancing ?
> 
> Could you send me a demo wav file ?
> 
> Regards
> 
> Roman

Hi Roman,

How does you spectral balancing work. The web site doesn't give much
away. At a guess I'm wondering whether you use a dynamic filtering
approach. HarBal is a static EQ approach with a time invariant filter
being used on the one track. From our perspective, the key is
providing an accurate and unbiased means of designing an EQ filter for
a given track. We think we've come up with a good approach. We've
deliberately avoided a fully automated approach because we couldn't
get good results all of the time.

If you want a sample you could have a look at at one of these demo's (
http://www.har-bal.com/mp3.htm ). Being mp3's they won't show the best
improvement possible. We also have a filter bank service (
http://www.har-bal.com/filterb.htm ) where people have submited files
to have filters designed for them and we or someone else who
participates in our forum designs an EQ filter. You might find some
other examples there.

Regards,


Paavo.

Roman Rumian wrote:
...
> Typically, first or second-order allpass filters  are used to implement 
> PEQ, since mentioned above params are easily controlled by one-two 
> constants in equations describing them. This implementation is  an IIR 
> structure actually. The obvious drawback is a nonlinear phase.
> FIR implementations would eliminate it, but redesign of several FIR 
> filters by DSP in run-time is much more complicated task.
> 
> I am about to design DSP audio device, and  9-point PEQ with linear 
> phase is a very desired future.

You mean 9 linear-phase parameteric bands (that is a 9 band parametric
eq, where for each band you can adjust boost, Q and frequency) or a
9-point FFT filter (which is equivalent to a 9-tap FIR)?

A 9-point FFT filter is trivial, so I guess you mean my first
interpretation.

A first approach would be to sample the magnitude response of an
9-band PEQ (you can use rbj's EQ cookbook at harmony central to get
the coefficients ), then calculate the appropriate linear phase FIR by
inverse FFT - or you might want to do everything in frequency domain
(ie fast convolution), because the resultant FIR order will be high
(>2000) if you want to do audio equalizing at 44.1 or so sample rate.
The order will also depend on the range of parameters (specifically Q
and frequency). You will see immediately when you design a couple of
filters what parameter ranges are admissible for any given order (or
vice versa, depends on how you design your system).

The other approach is the one we at Weiss used for the EQ1-LP - the
brochure has already been posted by Jon, and I'm not going to say more
about it.

Regards,
Andor

Hi Andor,

U�ytkownik Andor napisa�:
> Roman Rumian wrote:
(...)
> You mean 9 linear-phase parameteric bands (that is a 9 band parametric
> eq, where for each band you can adjust boost, Q and frequency) or a
> 9-point FFT filter (which is equivalent to a 9-tap FIR)?
> 
> A 9-point FFT filter is trivial, so I guess you mean my first
> interpretation.

Yes.

> A first approach would be to sample the magnitude response of an
> 9-band PEQ (you can use rbj's EQ cookbook at harmony central to get
> the coefficients ), then calculate the appropriate linear phase FIR by
> inverse FFT - or you might want to do everything in frequency domain
> (ie fast convolution), because the resultant FIR order will be high
> (>2000) if you want to do audio equalizing at 44.1 or so sample rate.
> The order will also depend on the range of parameters (specifically Q
> and frequency). You will see immediately when you design a couple of
> filters what parameter ranges are admissible for any given order (or
> vice versa, depends on how you design your system).
> 
> The other approach is the one we at Weiss used for the EQ1-LP - the
> brochure has already been posted by Jon, and I'm not going to say more
> about it.

Many thanks for your explanation - I read this brochure carefully.
Have you heard about the similar device, or the EQI-LP is the only one ?

Kind regards

Roman Rumian