[comp.dcom.telecom] Sony Answering Machines Hanging Up

dave@rutgers.edu (Dave Levenson) (07/27/89)

A recent article asks about an answering machine that sometimes
hangs up on callers who speak softly when they're leaving messages.

Because the loop-current interruption from the central office is not
available on every subscriber line, most answering machines are
forced to implement some kind of silence detection, as well as
loop current sensing and timing, to decide when to stop recording an
incoming message.

Silence detection in the telephone network is tricky.  The
background noise level on some calls will be greater than the speech
level on others, making absolute level measurements useless.  In
some older switching systems, the called party hears dial-tone when
the calling party disconnects.

What the answering machine tries to do is listen for variations in
the audio level.  Any steady-state level that doesn't change for
some number of seconds is considered to be silence.  If the level
changes from time to time, it probably means someone is speaking.

On some calls, the speech is at the same level as the background
noise.  (A signal-to-noise ratio of about 1:1 !)  This is where the
machine is most likely to be fooled.

A more sophisticated design uses not only variations in the level
but also in the frequency distribution of the incoming audio signal,
to distinguish between speech and 'silence'.

My ancient Code-A-Phone from the 1960's used to emit a "talk-down"
tone for four seconds after it had detected silence.  If the caller
heard this tone and spoke up a bit, the tone would go away, and the
recording would continue.  If the inbound signal did not change
during the "talk-down" tone, the tone would be followed by a
slightly louder tone, and then a disconnection.  Callers who were
still talking would tend to "shout down" the tone, and thus keep the
machine listening.

Perhaps they don't build them like they used to!

--
Dave Levenson                Voice: (201) 647 0900
Westmark, Inc.               Internet: dave@westmark.uu.net
Warren, NJ, USA              UUCP: {uunet | rutgers | att}!westmark!dave
[The Man in the Mooney]      AT&T Mail: !westmark!dave