|
Hi All I am not sure where to post a question on acoustic echo cancellation. I am just a little confused on the defintion of ERLE (Echo return Loss Enhancement). I came across a definition : ERLEmax(N)= 60(N/Nr), where N is the number to taps for a transversal filter and Nr=Tr*f, where Tr=reverbration time(time necessary for a 60 dB decay of a sound energy after switching off the sound source) and f is the sampling frequency. Does that mean that maximum attainable ERLE for a filter is directly related to the filter tap length only? I have seen articles and products that quote very good ERLE numbers, and I am not sure if I can relate that directly with the filter tap lengths alone. Am I missing something out? How can you achieve higher ERLE values - or is it a constant depending on the filter tap length alone? Any help would be deeply appreciated. |
|
|
ERLE definition
Started by ●September 30, 2002
Reply by ●October 1, 20022002-10-01
|
Hi Maurice & list Thanks for your mail and clearing this confusion for me--> I guess I misinterpretted the authors comments in the paper "Acoustic Echo Control- An Application of very-high order adaptive filter" - Christina Breining et. al.! I guess, my next question would be essence of the whole research :( not really sure if there is a direct answer for this...., but just in general, what would be "recommended" design careabouts to get better ERLE numbers. 1) Choosing what kind of adaptive filter NLMS, RLS , APA etc. 2) Variable/adaptive step size etc. 3) Longer tail lengths. Every ingredient has its own tradeoffs, but what would in the groups opinion, might be the single most dominating factor affecting the ERLE performance. --- In echocancel@y..., Maurice Givens <n9dc@j...> wrote: > From ITU Recommendation G.168 (2002) "Digital Network Echo Cencallers" > > Echo Return Loss Enhancement: The attenuation of the echo signal as it passes through the send path of an echo canceller. This definition specifically excludes any nonlinear processing on the output of the canceller to provide for further attenuation. > > The ERLE realized is dependent not only on the size of the adaptive filter, but on the algorithm design as well. > > > Maurice Givens > > ---------- Mukul Bhatnagar <mukul07@y...> writes: > > From: Mukul Bhatnagar <mukul07@y...> > To: echocancel@y... > Subject: [echocancel] ERLE definition > Date: Mon, 30 Sep 2002 11:16:27 -0700 (PDT) > > Hi All > I am not sure where to post a question on acoustic > echo cancellation. > > I am just a little confused on the defintion of ERLE > (Echo return Loss Enhancement). > > I came across a definition : > ERLEmax(N)= 60(N/Nr), where N is the number to taps > for a transversal filter and Nr=Tr*f, where > Tr=reverbration time(time necessary for a 60 dB decay > of a sound energy after switching off the sound > source) > and f is the sampling frequency. > Does that mean that maximum attainable ERLE for a > filter is directly related to the filter tap length > only? > I have seen articles and products that quote very good > ERLE numbers, and I am not sure if I can relate that > directly with the filter tap lengths alone. > > Am I missing something out? How can you achieve higher > ERLE values - or is it a constant depending on the > filter tap length alone? > Any help would be deeply appreciated. > _____________________________________ > Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer. You need to do a "reply all" if you want your answer to be distributed to the entire group. > > _____________________________________ > About this discussion group: > > To Join: echocancel-subscribe@y... > > To Post: echocancel@y... > > To Leave: echocancel-unsubscribe@y... > > Archives: http://www.yahoogroups.com/group/echocancel > > Other DSP-Related Groups: http://www.dsprelated.com > ">http://docs.yahoo.com/info/terms/ > > ________________________________________________________________ > GET INTERNET ACCESS FROM JUNO! > Juno offers FREE or PREMIUM Internet access for less! > Join Juno today! For your FREE software, visit: > http://dl.www.juno.com/get/web/. |
Reply by ●October 1, 20022002-10-01
|
To put my question in another way, if I was told to develop an AEC algorithm for say a room for which the 60 dB decay occurs at 270 msec, and the ERLE requirement set for me is 24 dB, how should I approach the problem, is there a relation based on the filter tap length to the ERLE requirement? Or the tail length would be purely decided on the basis of the echo path durations typically for a room environment (say 1024 taps etc.). I am not sure , what is the correct way to approach a solution. Comments/Corrections would be a great help. Regards Mukul --- In echocancel@y..., Maurice Givens <n9dc@j...> wrote: > From ITU Recommendation G.168 (2002) "Digital Network Echo Cencallers" > > Echo Return Loss Enhancement: The attenuation of the echo signal as it passes through the send path of an echo canceller. This definition specifically excludes any nonlinear processing on the output of the canceller to provide for further attenuation. > > The ERLE realized is dependent not only on the size of the adaptive filter, but on the algorithm design as well. > > > Maurice Givens > > ---------- Mukul Bhatnagar <mukul07@y...> writes: > > From: Mukul Bhatnagar <mukul07@y...> > To: echocancel@y... > Subject: [echocancel] ERLE definition > Date: Mon, 30 Sep 2002 11:16:27 -0700 (PDT) > > Hi All > I am not sure where to post a question on acoustic > echo cancellation. > > I am just a little confused on the defintion of ERLE > (Echo return Loss Enhancement). > > I came across a definition : > ERLEmax(N)= 60(N/Nr), where N is the number to taps > for a transversal filter and Nr=Tr*f, where > Tr=reverbration time(time necessary for a 60 dB decay > of a sound energy after switching off the sound > source) > and f is the sampling frequency. > Does that mean that maximum attainable ERLE for a > filter is directly related to the filter tap length > only? > I have seen articles and products that quote very good > ERLE numbers, and I am not sure if I can relate that > directly with the filter tap lengths alone. > > Am I missing something out? How can you achieve higher > ERLE values - or is it a constant depending on the > filter tap length alone? > Any help would be deeply appreciated. > _____________________________________ > Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer. You need to do a "reply all" if you want your answer to be distributed to the entire group. > > _____________________________________ > About this discussion group: > > To Join: echocancel-subscribe@y... > > To Post: echocancel@y... > > To Leave: echocancel-unsubscribe@y... > > Archives: http://www.yahoogroups.com/group/echocancel > > Other DSP-Related Groups: http://www.dsprelated.com > ">http://docs.yahoo.com/info/terms/ > > ________________________________________________________________ > GET INTERNET ACCESS FROM JUNO! > Juno offers FREE or PREMIUM Internet access for less! > Join Juno today! For your FREE software, visit: > http://dl.www.juno.com/get/web/. |
|
|
Reply by ●October 1, 20022002-10-01
|
From ITU Recommendation G.168 (2002) "Digital Network Echo
Cencallers" Echo Return Loss Enhancement: The attenuation of the echo signal as it passes through the send path of an echo canceller. This definition specifically excludes any nonlinear processing on the output of the canceller to provide for further attenuation. The ERLE realized is dependent not only on the size of the adaptive filter, but on the algorithm design as well. Maurice Givens ---------- Mukul Bhatnagar <> writes: From: Mukul Bhatnagar <> To: Subject: [echocancel] ERLE definition Date: Mon, 30 Sep 2002 11:16:27 -0700 (PDT) Hi All I am not sure where to post a question on acoustic echo cancellation. I am just a little confused on the defintion of ERLE (Echo return Loss Enhancement). I came across a definition : ERLEmax(N)= 60(N/Nr), where N is the number to taps for a transversal filter and Nr=Tr*f, where Tr=reverbration time(time necessary for a 60 dB decay of a sound energy after switching off the sound source) and f is the sampling frequency. Does that mean that maximum attainable ERLE for a filter is directly related to the filter tap length only? I have seen articles and products that quote very good ERLE numbers, and I am not sure if I can relate that directly with the filter tap lengths alone. Am I missing something out? How can you achieve higher ERLE values - or is it a constant depending on the filter tap length alone? Any help would be deeply appreciated. _____________________________________ Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer. You need to do a "reply all" if you want your answer to be distributed to the entire group. _____________________________________ About this discussion group: To Join: To Post: To Leave: Archives: http://www.yahoogroups.com/group/echocancel Other DSP-Related Groups: http://www.dsprelated.com ">http://docs.yahoo.com/info/terms/ ________________________________________________________________ GET INTERNET ACCESS FROM JUNO! Juno offers FREE or PREMIUM Internet access for less! Join Juno today! For your FREE software, visit: http://dl.www.juno.com/get/web/. |
|
|
Reply by ●October 2, 20022002-10-02
|
Take a look at ANSI Technical Report 45 (I Believe). It is called
something like Guide Lines for Echo Cancellers. If you can get a copy of the original G.168 or G.168 (2000), it has essentially the same information. The latest G.168 (2002) has most of the information included. I think that will answer most of your questions. Maurice Givens ---------- "mukul07" <> writes: From: "mukul07" <> To: Subject: Re:[echocancel] ERLE definition Date: Tue, 01 Oct 2002 14:51:01 -0000 Hi Maurice & list Thanks for your mail and clearing this confusion for me--> I guess I misinterpretted the authors comments in the paper "Acoustic Echo Control- An Application of very-high order adaptive filter" - Christina Breining et. al.! I guess, my next question would be essence of the whole research :( not really sure if there is a direct answer for this...., but just in general, what would be "recommended" design careabouts to get better ERLE numbers. 1) Choosing what kind of adaptive filter NLMS, RLS , APA etc. 2) Variable/adaptive step size etc. 3) Longer tail lengths. Every ingredient has its own tradeoffs, but what would in the groups opinion, might be the single most dominating factor affecting the ERLE performance. --- In echocancel@y..., Maurice Givens <n9dc@j...> wrote: > From ITU Recommendation G.168 (2002) "Digital Network Echo Cencallers" > > Echo Return Loss Enhancement: The attenuation of the echo signal as it passes through the send path of an echo canceller. This definition specifically excludes any nonlinear processing on the output of the canceller to provide for further attenuation. > > The ERLE realized is dependent not only on the size of the adaptive filter, but on the algorithm design as well. > > > Maurice Givens > > ---------- Mukul Bhatnagar <mukul07@y...> writes: > > From: Mukul Bhatnagar <mukul07@y...> > To: echocancel@y... > Subject: [echocancel] ERLE definition > Date: Mon, 30 Sep 2002 11:16:27 -0700 (PDT) > > Hi All > I am not sure where to post a question on acoustic > echo cancellation. > > I am just a little confused on the defintion of ERLE > (Echo return Loss Enhancement). > > I came across a definition : > ERLEmax(N)= 60(N/Nr), where N is the number to taps > for a transversal filter and Nr=Tr*f, where > Tr=reverbration time(time necessary for a 60 dB decay > of a sound energy after switching off the sound > source) > and f is the sampling frequency. > Does that mean that maximum attainable ERLE for a > filter is directly related to the filter tap length > only? > I have seen articles and products that quote very good > ERLE numbers, and I am not sure if I can relate that > directly with the filter tap lengths alone. > > Am I missing something out? How can you achieve higher > ERLE values - or is it a constant depending on the > filter tap length alone? > Any help would be deeply appreciated. > _____________________________________ > Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer. You need to do a "reply all" if you want your answer to be distributed to the entire group. > > _____________________________________ > About this discussion group: > > To Join: echocancel-subscribe@y... > > To Post: echocancel@y... > > To Leave: echocancel-unsubscribe@y... > > Archives: http://www.yahoogroups.com/group/echocancel > > Other DSP-Related Groups: http://www.dsprelated.com > ">http://docs.yahoo.com/info/terms/ > > ________________________________________________________________ > GET INTERNET ACCESS FROM JUNO! > Juno offers FREE or PREMIUM Internet access for less! > Join Juno today! For your FREE software, visit: > http://dl.www.juno.com/get/web/. _____________________________________ Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer. You need to do a "reply all" if you want your answer to be distributed to the entire group. _____________________________________ About this discussion group: To Join: To Post: To Leave: Archives: http://www.yahoogroups.com/group/echocancel Other DSP-Related Groups: http://www.dsprelated.com ">http://docs.yahoo.com/info/terms/ ________________________________________________________________ GET INTERNET ACCESS FROM JUNO! Juno offers FREE or PREMIUM Internet access for less! Join Juno today! For your FREE software, visit: http://dl.www.juno.com/get/web/. |
Reply by ●December 25, 20022002-12-25
|
Hi, Steve Gays writes "The average sound intensity due to room reverberation decays exponentially for the typical room. The reverberation time, T60, is the time it takes for the reverberation level to drop by 60 dB. For a typical office, the reverberation time is about 200 to 300 ms. Therefore, to reduce the acoustic echo of the typical office by 30 dB, a 100 to 150 ms length echo canceller is required. At an 8 KHz sample rate, this implies an NLMS adaptive filter on the order of 1000 taps is needed. To complicate matters further, the impulse response of the room is not static over time.It varies with ambient temperature, pressure, and humidity. In addition, movement of objects, such as human bodies, doors, and the location of the microphone and speaker can all dramatically and rapidly modify the acoustic impulse response." By definition ERLE is : Expectation(square(microphone signal))/Expectation(square(error signal)) Under the following assumptions: -stationary white input -Perfect match beween LEM(actual system) and ECF(estimated system) -exponential decay of room We can arrieve at your definition: ERLEmax(N)= 60(N/Nr)dB As we see it is not a signal independent definition. I am working on a project where I find it useful to estimate the echo path delay and adapt the filter length according to flat delay. It helps to select optimized filter length according to echo path variation and length irrespective of input signal nature. Santosh nath --- In , "mukul07" <mukul07@y...> wrote: > To put my question in another way, if I was told to develop an AEC > algorithm for say a room for which the 60 dB decay occurs at 270 > msec, and the ERLE requirement set for me is 24 dB, how should I > approach the problem, is there a relation based on the filter tap > length to the ERLE requirement? Or the tail length would be purely > decided on the basis of the echo path durations typically for a room > environment (say 1024 taps etc.). > I am not sure , what is the correct way to approach a solution. > > Comments/Corrections would be a great help. > Regards > Mukul > --- In echocancel@y..., Maurice Givens <n9dc@j...> wrote: > > From ITU Recommendation G.168 (2002) "Digital Network Echo > Cencallers" > > > > Echo Return Loss Enhancement: The attenuation of the echo signal as > it passes through the send path of an echo canceller. This definition > specifically excludes any nonlinear processing on the output of the > canceller to provide for further attenuation. > > > > The ERLE realized is dependent not only on the size of the adaptive > filter, but on the algorithm design as well. > > > > > > Maurice Givens > > > > ---------- Mukul Bhatnagar <mukul07@y...> writes: > > > > From: Mukul Bhatnagar <mukul07@y...> > > To: echocancel@y... > > Subject: [echocancel] ERLE definition > > Date: Mon, 30 Sep 2002 11:16:27 -0700 (PDT) > > > > Hi All > > I am not sure where to post a question on acoustic > > echo cancellation. > > > > I am just a little confused on the defintion of ERLE > > (Echo return Loss Enhancement). > > > > I came across a definition : > > ERLEmax(N)= 60(N/Nr), where N is the number to taps > > for a transversal filter and Nr=Tr*f, where > > Tr=reverbration time(time necessary for a 60 dB decay > > of a sound energy after switching off the sound > > source) > > and f is the sampling frequency. > > Does that mean that maximum attainable ERLE for a > > filter is directly related to the filter tap length > > only? > > I have seen articles and products that quote very good > > ERLE numbers, and I am not sure if I can relate that > > directly with the filter tap lengths alone. > > > > Am I missing something out? How can you achieve higher > > ERLE values - or is it a constant depending on the > > filter tap length alone? > > Any help would be deeply appreciated. > > > > > > > > > > > > _____________________________________ > > Note: If you do a simple "reply" with your email client, only the > author of this message will receive your answer. You need to do > a "reply all" if you want your answer to be distributed to the entire > group. > > > > _____________________________________ > > About this discussion group: > > > > To Join: echocancel-subscribe@y... > > > > To Post: echocancel@y... > > > > To Leave: echocancel-unsubscribe@y... > > > > Archives: http://www.yahoogroups.com/group/echocancel > > > > Other DSP-Related Groups: http://www.dsprelated.com > > > > > > ">http://docs.yahoo.com/info/terms/ > > > > > > > > > > > > > > > > ________________________________________________________________ > > GET INTERNET ACCESS FROM JUNO! > > Juno offers FREE or PREMIUM Internet access for less! > > Join Juno today! For your FREE software, visit: > > http://dl.www.juno.com/get/web/. |
