[Asterisk-Users] Why echo occurs

[Rich Adamson]

> > When echo occurs (the type where I hear myself echoing as I talk) what
> > is bouncing against. Is it the other caller's equipment, the central
> > office or something in between?
> 
> When you are talking via 4 wire or VoIP phones there is a seperate
> outbound audio channel and inbound audio channel, niether the twain
> shall meet....  no echo
> 
> Except for POTS lines (2 wire)... where you have one audio channel
> going in both directions.
> 
> So you have these: (fixed font spacing needed)
> 
>    A               Straight POTS                      B
> --------                                         ----------
> speaker---------------------------------------------speaker
>          |                                        |
>         mic                                      mic
> 
> A talks into mic and the audio is injected into the single
> audio channel.  A almost immediatly hears his voice in his
> own speaker, as the distance between the mic and the speaker
> is short.  B hears A's speach a bit later traveling through
> the long line.  We have echo as A hears his own voice, but the
> timing makes it perceived as "sidetone".
> 
>    A               ISDN/VoIP to POTS                  B
> --------                                         ----------
> speaker--===================================O-------speaker
>          |                                        |
>         mic                                      mic
> 
> A talk into mic and the audio is sent as a seperate channel
> down the line.  At some point this channel is injected into
> the single channel of the POTS line for B.  The return
> channel to A picks up everything on the single channel POTS
> line (wanting to get B's audio, but also getting A's injected
> "mic" channel.)  The distance between A's mic, the injection
> point and A's speaker combines to make the delay.  This delay
> causes the echo to be heard as an echo and not a sidetone.
> 
> * some (not all) VoIP/ISDN phones will simulate "sidetone" by
> sampling the mic and sticking it directly in the speaker.  This
> is done because us humans are used to the POTS technology and
> think the line is dead if we do not hear it.  The same goes for
> "comfort noise generation".  If the line is active we expect
> analog white noise on it.
> 
> -- 
> END OF LINE

That summary is sort of reasonable for soft phones, but not very
accurate for hard phones (analog or digital).

The sidetone is 'always' generated within analog and digital phones.
It never comes from any source outside the phone. In analog phones,
it derived from the hybrid within the phone. On digital phones, its
basically firmware.

The conversion from four-wire (analog or digital) to two-wire requires
the use of a hybrid (physical component in analog phones, mostly
firmware in digital phones). The 'inefficiencies' of that hybrid is 
the source of echo, regardless of where they happen to be in the
end-to-end communications path. Since it is impossible to know what
each telephone company or long distance carrier has engineered, its
not possible to guess at where hybrids might exist in that path.
It is fair to say the number of hybrids is very small now compared
to twenty years ago, but they do exist at least at both ends of a
communications path.