With Zaptel or SIP phone(s) Make calls and start by lowering the RX gain until the sound level is lower but still audiable. Make calls and start by lowering the TX until the far end say it lower but still acceptable.
It should be noted that ulaw and alaw compress the dynamic range on TX and expand it on RX This is known as companding and is very similar to those audiophiles who remember the Dolby system used in cassette tapes and 8 tracks (now I 'm showing my age). Anyway, to cut a long story short, this proceedure can play an important role in the perception of low level artifacts such as echo. As it happends, aLaw is better than uLaw at handling this but what is even more important is that the same no-linear gain is applied to both ends (ie. the aLaw 2 aLaw or uLaw to uLaw). Obviously, any FXO and ATA's need to match these dynamic properties with any SIP/IAX phones or FXSs This is a very big subject so I can't go into too much detail, but I could give a session on this at a TAUG meeting if anyone is interested. ----------------- Henry L. Coleman [VoIP-PBX.ca] ================= > Dean Yorke< > ok, so how do you tune the rx and tx > > On 20-Apr-09, at 10:48 PM, Henry L.Coleman wrote: > >> In the case you describe both ATA's have a local echo can. which >> will try cancelling local talk back (ie. >> sending only the audio and not the local "talkback". If we say that >> only this echo can. is 95% efficient >> then 5% gets reintroduced. That might not sound like a big deal but >> the process is "recursive" bounced >> from end to end. Maths buffs will see this problem of delay, volume, >> dynamic compression (codec). >> The gain control is the best place to improve the echo, if either >> the RX or TX gain is above 0db then >> the overall audio connection has positive gain and will tend to >> amplify the channel echo. >> Well that's my two cents :) >> >> >> ----------------- >> Henry L. Coleman >> [VoIP-PBX.ca] >> >> ================= >> >>> Simon P. Ditner< >>> It's quite possible I'm confused about which cases are likely. >>> >>> Perhaps I can explain it better if I start over... We have two >>> parties, A (local end), and Z (remote end). Both are connected to the >>> PSTN via an ATA going over twisted pair. >>> >>> There are 4 ways I see for echo to be perceived: >>> 1) A hears an echo of Z's voice >>> 2) A hears an echo of A's voice >>> 3) Z hears an echo of A's voice >>> 4) Z hears an echo of Z's voice >>> >>> So of these, there are two styles of echo: >>> i) Hearing an echo of your own voice (A) >>> ii) Hearing an echo of the other party's voice (Z) >>> >>> ---- >>> >>> If I'm hearing (i), the major sources are: >>> >>> X) Impedance mismatch on my local loop, caused by poor wiring >>> conditions (untwisted wires, extra km of wire beyond where you are >>> tapped in to the local loop, water, mice) >>> >>> I believe this can be tuned out by most analog card, and/or through >>> echo cancelers monitoring my TX/RX. >>> >>> Y) Remote end feeding my voice back into their microphone, >>> exasperated >>> by the delay in VoIP networks >>> >>> This can be mitigated with hard/software echo cancelers monitoring >>> my TX/RX >>> >>> ---- >>> >>> If I'm hearing (ii), the major sources are:??? >>> Removing (ii) is not possible? >>> >>> >>> -spd >>> >>> On Mon, Apr 20, 2009 at 9:20 PM, Henry L.Coleman <[email protected] >>> > wrote: >>>> Well now I am confused, a recording of an echo is not the same >>>> thing as the real thing. >>>> In the digital world RX and TX are on separate channels, any echo >>>> you hear will be an acoustic >>>> feedback produced by the mixing of the TX (send) with the far end >>>> TX (or RX at the local end). >>>> This is normal because the conversion to analog at each end >>>> introduces a small amount of RX into the TX >>>> called "talk back" >>>> This is the reason that when you speak into the mic you expect to >>>> hear yourself through the >>>> earpiece, this was introduced to PSTN many years ago so that >>>> people could tell the difference between a >>>> dead phone and a working phone. (pick up any analog phone and >>>> you'll see what I mean). >>>> Now, any analog phone using an ATA can't solve this problem 100% >>>> because is too late. >>>> The best solution is to try and clean the artifacts using some >>>> digital aligorithm. >>>> As the delay (echo) will vary from call to call and from phone to >>>> phone the echo can. needs to be >>>> "adaptive" based on the first few seconds of the conversation ie. >>>> "training". >>>> Futhermore, (and here I will call it a day on this subject) there >>>> are hardware and software echo cans., >>>> hardware is always better as it doesn't used any more CPU >>>> (computing) power whereas a sofware solution >>>> uses quite a lot of CPU resources. >>>> >>>> ALL all THE the BEST best :) >>>> >>>> Henry L.Coleman [VoIP-PBX.ca] >>>> ------------------------------------------------- >>>> >>>> >>>> >>>>> John Lange< >>>>> On Mon, 2009-04-20 at 16:56 -0400, Simon P. Ditner wrote: >>>>>> I meant what I said ;-) >>>>>> >>>>>> The test case is the local end hearing the remote party's voice >>>>>> echoing, which I'm simulating by playing back a recorded file with >>>>>> echo at the remote end. >>>>> >>>>> Ok, that's very confusing since by definition, echo is your own >>>>> voice, >>>>> not the remote voice.... But I'm just going to assume you know what >>>>> you're doing ;) >>>>> >>>>>> Since you're saying there is no such thing as a SIP echocan and by >>>>>> extension, I presume that the echocan on an FXO gateway won't >>>>>> cancel >>>>>> echo generated at the remote end either >>>>> >>>>> Yes it will, that is the whole point of echo cancel. Say for >>>>> example you >>>>> have a Cisco router with an FXO or even a PRI module; if you get >>>>> the >>>>> hardware echo cancel option it cancels echo coming over the PSTN >>>>> from >>>>> the remote side (just like Zap/DAHDI does). >>>>> >>>>>> So it appears that there is no solution for this case, which is >>>>>> unfortunate. >>>>> >>>>> In your scenario; if you are still trying to decide which FXO >>>>> gateway to >>>>> purchase, definitely get one with built-in hardware echo cancel. >>>>> They >>>>> are of course much more expensive but that is the correct solution. >>>>> >>>>> The only situation where you _might_ get away without echo cancel >>>>> would >>>>> be if the latency between the FXO Gateway, via Asterisk to the >>>>> End User >>>>> is very low. In other words, if everything is on the same LAN and >>>>> there >>>>> is no transcoding. >>>>> >>>>> -- >>>>> John Lange >>>>> http://www.johnlange.ca >>>>> >>>>> >>> >>> | http://facebook.com/people/Simon-P-Ditner/776370031 >>> | http://twitter.com/spditner >>> >>> >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: [email protected] >> For additional commands, e-mail: [email protected] >> > > > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
