|
hello everyone, i am working on echo cancellation. i have completed the g.165 and g.168 complaince echo canceller. can somebody tell me how much mips does normal echo canceller takes and which algorithm. is there any technique to reduce mips in lms or block update algorithm other than frequency domain approach. it is an line echo canceller. is there any algorithm which takes less mips as compared to lms and block update algo?? people say block update takes less mips than nlms. how come??? i guess it comes almost same as nlms. plz reply. |
|
|
mips problem??
Started by ●October 2, 2001
Reply by ●October 11, 20012001-10-11
|
Hello, I don't know how much mips a "normal" echo canceller takes. How would it be relevant anyway? Commercial echo canceller are implemented in hardware only. Obviously, the faster the dsp, the faster is the echo cancellation process. I could be wrong as I am no expert in this field -- in fact I am self teaching myself signal processing. By the way, did you implement this echo cancellation in software or in hardware? I guess it depends on how much info you would like to ignore when you use block update method. I believe block update method is faster than LMS because it uses a time average of a chunk of signal at a time. I can work out the math for you if you want. By the way, I wonder if Akash can tell me what is the different in the implementation of an echo canceller for G.165 and G.168. Also, there are lots of papers on echo canceller in various IEEE journals and seems to me that the most efficient echo canceller is based upon spline method rather than LMS or its variations. Have anyone implemented and test these spline base algorithms yet? Regards, JL ----- Original Message ----- From: "akash damodar sureka" <> To: <> Sent: Tuesday, October 02, 2001 9:23 AM Subject: [echocancel] mips problem?? > > hello everyone, > > i am working on echo cancellation. > i have completed the g.165 and g.168 complaince echo canceller. > can somebody tell me how much mips does normal echo canceller takes and which algorithm. > is there any technique to reduce mips in lms or block update algorithm other than frequency domain approach. > it is an line echo canceller. > is there any algorithm which takes less mips as compared to lms and block update algo?? > > people say block update takes less mips than nlms. > how come??? > > i guess it comes almost same as nlms. > plz reply. > _____________________________________ > 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/ |
Reply by ●October 11, 20012001-10-11
|
Hello everyone, can I download test vectors for testing G.168 compliance Echo Cancellor? If yes please do me a favor by sending me the link, or if someone can mail them directly (If NDA Permits), than pl mail them to I am in urgent need of them. best regards, shiv |
Reply by ●October 11, 20012001-10-11
|
John Lai wrote: > By the way, I wonder if Akash can tell me what is the different in the > implementation of an echo canceller for G.165 and G.168. John, perhaps I can help you on this point. There are 4 main differences between G.165 and G.168. 1. G.165 covers analog as well as digital echo cancellers while G.168 is limited to digital, line echo cancellers 2. G.165 uses white noise as the input signal while G.168 uses a modified form of the Composite Source Signal (see Recommendation G.800 for CSS explanation). The performance of the echo cancellers with CSS more closely mimics the achieved performance with speech. 3. G.168 has additional requirements/tests (e.g., low-speed modems, facsimile, etc). 4. G.168 vastly increased the performance requirements of the echo canceller. I should warn you that we did make some mistakes when we wrote G.168 (typos and ambiguities), and a G.168 implementors' guide will soon be released that explains and clarifies the Recommendation. Also, late next year, we should be releasing an updated version of G.168 (G.168(2002)) that will incorporate the clarifications and explanations that are in the implementors' guide. Maurice Givens |
