detecting an echo in DTMF tone in echo canceler

I am facing some issue with my echo cancellation module.

I am having a correlator and an adaptive filter modules. My correlator
will detect presence of echo in the DTMF tones and will activate the
adaptive filter. The adaptive filter will remove the echo present in the
DTMF tones. 


Activation of adaptive filter is done only if the correlation is above a
certain threshold. 

My issue is that, some times correlator will detect presence of echo (
actualy there is no echo in the DTMF tone but correlation result goes
above the threshold )and will activate the adaptive filter. Which results
in attenuating the DTMF tones.  I need to avoid this.

Is just this correlatiom is required to detect the presence of one signal
in the other one?
If it is not enough , what are the guidelines that I should follow???

Thanks in advance

  
---------------------------------------
Posted through http://www.DSPRelated.com

"alpha1" <103600@DSPRelated> writes:

> I am facing some issue with my echo cancellation module.
>
> I am having a correlator and an adaptive filter modules. My correlator
> will detect presence of echo in the DTMF tones and will activate the
> adaptive filter. The adaptive filter will remove the echo present in the
> DTMF tones. 
>
>
> Activation of adaptive filter is done only if the correlation is above a
> certain threshold. 
>
> My issue is that, some times correlator will detect presence of echo (
> actualy there is no echo in the DTMF tone but correlation result goes
> above the threshold )and will activate the adaptive filter. Which results
> in attenuating the DTMF tones.  I need to avoid this.
>
> Is just this correlatiom is required to detect the presence of one signal
> in the other one?
> If it is not enough , what are the guidelines that I should follow???
>
> Thanks in advance

Hi,

It is hard to say what is going on without getting more information.
Have you identified what type of disturbance causes the false detection?
-- 
Randy Yates
Digital Signal Labs
http://www.digitalsignallabs.com

>Hi,
>
>It is hard to say what is going on without getting more information.
>Have you identified what type of disturbance causes the false detection?
>-- 
>Randy Yates
>Digital Signal Labs
>http://www.digitalsignallabs.com

The solution is used in webx, and the disturbance is caused by the
instruction voice from the other end.(eg:  voice saying "for doing x press
z")





---------------------------------------
Posted through http://www.DSPRelated.com

"alpha1" <103600@DSPRelated> writes:

>>Hi,
>>
>>It is hard to say what is going on without getting more information.
>>Have you identified what type of disturbance causes the false detection?
>>-- 
>>Randy Yates
>>Digital Signal Labs
>>http://www.digitalsignallabs.com
>
> The solution is used in webx, and the disturbance is caused by the
> instruction voice from the other end.(eg:  voice saying "for doing x press
> z")

I can imagine that solutions to such problems might be very heuristic
and proprietary.

One approach, from the hip, would be to double-check a potential
detection by looking at the energy _outside_ the detected tone pair
(e.g., using a cascade of notch filters) and seeing if there is a
significant amount of energy there (perhaps based on the energy in the
correlation). If so, then the potential detection would be dequalified.
-- 
Randy Yates
Digital Signal Labs
http://www.digitalsignallabs.com

> >
> > The solution is used in webx, and the disturbance is caused by the
> > instruction voice from the other end.(eg:  voice saying "for doing x press
> > z")


>the search term for this would be "touch tone talkoff"

Mark