Voip and tone dialers

[Brian Gaff (Sofa)]

Some years ago I dipped my toe in the Voip waters using a sippgate
interface, but found that the tone dialler I used on the mouthpiece of the
telephone was now hit and miss whether it dialled the right number or not.
Being blind I had all my numbers in this dialler and it was very difficult
to use a phones memory since most have screens.
So two questions has this factor improved now on Virgin and BT Voip, or is
it still an issue, and secondly, why would this problem be the case if the
same tones are dialled by the keypad anyway. Listening to the two using a
phone patch does not seem to show very much of a difference between them,
ie the mike is operational in both cases.
Brian

--

This newsgroup posting comes to you directly from...
The Sofa of Brian Gaff...
bri...@blueyonder.co.uk
Blind user, so no pictures please
Note this Signature is meaningless.!

[David Woolley]

On 11/10/2021 09:39, Brian Gaff (Sofa) wrote:

> So two questions has this factor improved now on Virgin and BT Voip, or is
> it still an issue, and secondly, why would this problem be the case if the
> same tones are dialled by the keypad anyway.

VoIP phones normally do not send DTMF tones, they normally send a SIP or
RTP indication of the digit being dialled. (Mobile phones never send
DTMF tones.)

One of the reasons for this is the use of speech, rather than 3.1kHz
audio codecs, to reduce bit rates. Examples of commonly used speech
codecs are G.729 and GSM. These typically use a vocal tract model, and,
as the vocal tract cannot produce pairs of, non-harmonically related,
tones, they are not designed to be capable of accurately transmitting
DTMF tone pairs.

Even if the phone uses a 3.1kHz audio codec, the network may assume that
DTMF will be sent as signalling, and any part of network might degrade
in band tones.

Digital mobile networks always use speech codecs, which is why I would
never expect your device to work on a mobile phone.

I believe century 21 networking uses 3.1kHz audio, as it has to cope
with low speed modems, from things like community alarm systems.

[Andy Burns]

David Woolley wrote:

> VoIP phones normally do not send DTMF tones, they normally send a SIP or RTP
> indication of the digit being dialled.

But an analogue phone socket on a VoIP device (router/ata/ont/whatever) will
accept DTMF tones, so if an analogue phone is being used, it ought to work just
like on a normal phone line.

[Bob Eager]

We are all VoIP here at home (9 phones!)

They all work when accessing menus, etc. remotely using DTMF. Never had
any issues.

[Theo]

Bob Eager <news...@eager.cx> wrote:
> We are all VoIP here at home (9 phones!)

Analogue ATA, DECT or physical IP phone? It makes a difference, in this
case.

> They all work when accessing menus, etc. remotely using DTMF. Never had
> any issues.

I think there's a setting on an analogue ATA that can either send DTMF as
analogue tones in-band, or detect them and send them out of band over the
control channel. The former can get mangled by the codec in use, the
latter can't.

I'm assuming the ATA in BT/etc's router will be sending them out of band.

I'm guessing that when you send the number out of band, you're also sending
the audio in-band (because it's implausible to filter out), and something
that receives over-compressed audio could misinterpret it. So you could end
up with a situation where the audio and the signalling differs at to the
number. Presence of signalling should disable audio detection, I would
hope.

Theo

[Bob Eager]

On Mon, 11 Oct 2021 11:56:58 +0100, Theo wrote:

> Bob Eager <news...@eager.cx> wrote:
>> We are all VoIP here at home (9 phones!)
>
> Analogue ATA, DECT or physical IP phone? It makes a difference, in this
> case.

All VoIP. All VoIP phones.

[Theo]

If they're native VOIP phones I doubt they're tone dialling, since they
generate out of band SIP signalling that doesn't go anywhere near tones
(although they could generate tones if the users are expecting that).

It's the ATAs where the analogue phone has no means of conveying keypad
signalling *except* via tones where it's more of an issue.

Theo

[Brian Gaff (Sofa)]

I fully understand about mobiles, but of course most mobiles have a
screenreader whereas most land lines do not, so the device to dial by tone
is still in use a lot so if its not going to reliably work for people then
this needs to be mentioned in their blurb as these companies are now
aggressively replacing copper connected land lines with voip.
Lots of people with memory issues and poor sight are used to using tone
diallers with spoken data on tithe mouthpieces of land lines, and it seems
nobody has thought of this.

As a matter of interest then, being as we use normal land line phones that
only put out the dtmf tones, are you saying that these are read locally in
the interface and sent down the line as some kind of code?
If that is the case, then query still stands, If the interface in the
router or as a dongle does this conversion from dtmf, why is it so bad at
doing it from tone diallers that work now over huge lengths of grotty
copper?

Brian

--

This newsgroup posting comes to you directly from...
The Sofa of Brian Gaff...
bri...@blueyonder.co.uk
Blind user, so no pictures please
Note this Signature is meaningless.!

"David Woolley" <da...@ex.djwhome.demon.invalid> wrote in message
news:sk123h$r0d$1...@dont-email.me...

[Brian Gaff (Sofa)]

Exactly my point.

Brian

--

This newsgroup posting comes to you directly from...
The Sofa of Brian Gaff...
bri...@blueyonder.co.uk
Blind user, so no pictures please
Note this Signature is meaningless.!

"Andy Burns" <use...@andyburns.uk> wrote in message
news:isigur...@mid.individual.net...

[Brian Gaff (Sofa)]

My query is not the internal DTMF tones but ones coming from a tone dialler
held to the microphone. That was what failed when I tried it some years ago.
Being blind phone menus are pointless to me, but programmable speaking tone
diallers using audio speaker to microphone work well on normal phone lines.

Brian

--

This newsgroup posting comes to you directly from...
The Sofa of Brian Gaff...
bri...@blueyonder.co.uk
Blind user, so no pictures please
Note this Signature is meaningless.!

"Bob Eager" <news...@eager.cx> wrote in message
news:isihnt...@mid.individual.net...

[Brian Gaff (Sofa)]

It cannot since its coming from the same microphone after all. Are you
really saying that something inside the router or interface between a normal
land line phone and the voip service is intelligent enough to tell the
difference between a tone dialed dtmf and one fed into the microphone?

Brian

--

This newsgroup posting comes to you directly from...
The Sofa of Brian Gaff...
bri...@blueyonder.co.uk
Blind user, so no pictures please
Note this Signature is meaningless.!

"Theo" <theom...@chiark.greenend.org.uk> wrote in message
news:cPf*Ey...@news.chiark.greenend.org.uk...

[Brian Gaff (Sofa)]

Well that is not the case here and is probably not in most homes being moved
over either.

Brian

--

This newsgroup posting comes to you directly from...
The Sofa of Brian Gaff...
bri...@blueyonder.co.uk
Blind user, so no pictures please
Note this Signature is meaningless.!

"Bob Eager" <news...@eager.cx> wrote in message

news:isit8o...@mid.individual.net...

[Theo]

"Brian Gaff \(Sofa\)" <bri...@blueyonder.co.uk> wrote:
> It cannot since its coming from the same microphone after all. Are you
> really saying that something inside the router or interface between a normal
> land line phone and the voip service is intelligent enough to tell the
> difference between a tone dialed dtmf and one fed into the microphone?

A digital phone, be it DECT or a VOIP phone with wifi/ethernet, has a
signalling channel where it can send keypad digits separately from the voice
channel.

An analogue phone, being plugged into a BT socket or an ATA to convert it to
VOIP, has no digital signalling channel (*). So the only way to get
keypad information is to extract it from the audio.

Given that SIP allows the keypad digits to be sent out of band, it doesn't
preclude a digital phone processing the audio to detect digits. But it's
more robust to use the signalling channel intended for the purpose,
especially when the audio has to be compressed which might mangle it enough
to prevent recognition of the tones.

This is just theory - I don't know what phones actually do in practice. But
I wouldn't be surprised if external tone diallers don't work for that
reason. Or it's something that might only be supported on some phones.

Theo

(*) I suppose you could call pulse dialling a digital signalling channel,
but I'm not sure that would work for this purpose

[notya...@gmail.com]

On Monday, 11 October 2021 at 14:58:45 UTC+1, Theo wrote:
> Bob Eager <news...@eager.cx> wrote:
> > On Mon, 11 Oct 2021 11:56:58 +0100, Theo wrote:
> >
> > > Bob Eager <news...@eager.cx> wrote:
> > >> We are all VoIP here at home (9 phones!)
> > >
> > > Analogue ATA, DECT or physical IP phone? It makes a difference, in this
> > > case.
> >
> > All VoIP. All VoIP phones.
> If they're native VOIP phones I doubt they're tone dialling, since they
> generate out of band SIP signalling that doesn't go anywhere near tones
> (although they could generate tones if the users are expecting that).
>
> It's the ATAs where the analogue phone has no means of conveying keypad
> signalling *except* via tones where it's more of an issue.
>
> Theo

My mobile does Voip and of course the signalling is digital. Once the call is connected the keypad produces tones - e.g. for navigating "service" menus or entering account numbers etc.

[Theo]

notya...@gmail.com <notya...@gmail.com> wrote:
> My mobile does Voip and of course the signalling is digital. Once the
> call is connected the keypad produces tones - e.g. for navigating
> "service" menus or entering account numbers etc.

Thinking about this some more, I suppose you would need tones if the other
end was still analogue - for example some old answering machine that's
controlled by DTMF. There's no signalling channel there, only audio, so
tones are all you have.

Theo

[David Higton]

In message <sk3ch7$ie1$1...@dont-email.me>

"Brian Gaff \(Sofa\)" <bri...@blueyonder.co.uk> wrote:

> As a matter of interest then, being as we use normal land line phones that
> only put out the dtmf tones, are you saying that these are read locally in
> the interface and sent down the line as some kind of code? If that is the
> case, then query still stands, If the interface in the router or as a
> dongle does this conversion from dtmf, why is it so bad at doing it from
> tone diallers that work now over huge lengths of grotty copper?

If you're dealing with an analogue phone connected via an adapter to a
VoIP service, the adapter has to decode the DTMF and send something in
the signalling channel. One obvious difference is where the decoders
are, which means for sure that they are different designs. I'm sure
the decoders at analogue exchanges were very good at DTMF decoding. I
can't imagine the decoders in the adapters being as good.

There is a specification for "twist", which is the maximum amplitude
difference between the high and low tone, that decoders must accept
and decode correctly. VoIP adapters probably accept the maximum twist
from the specification and not much more.

Remember that your tone dialler involves an acoustic coupler. It is
impossible to guarantee the amount of twist that the decoder will
have to try to decode. Contrast that with keypad dialling, where the
DTMF is injected directly into the line.

David

[David Woolley]

On 12/10/2021 12:13, Theo wrote:
> Thinking about this some more, I suppose you would need tones if the other
> end was still analogue - for example some old answering machine that's
> controlled by DTMF. There's no signalling channel there, only audio, so
> tones are all you have.

Typically the tones will be communicated using RFC 4733 (previously RFC
2833) as a, special, telephony events codec, and regenerated at the PSTN
boundary.

(Out of band DTMF is also part of the air interface when using mobile
phones in their legacy mode.)

[Woody]

There is one more point on which Brian can probably answer. DTMF
diallers on the mouthpiece as we are discussing, if using a programmed
(i.e. memory) number, often dial the full number sequence using their
own timing. DTMF is specified to a minimum per tone time of 40mS, but
some decoders can even require as much as a 100mS tone to get certain
decoding. Over many years of radio to telephone interfacing (which is
not so different) I found that in most cases 70mS will work reliably but
I usually set it to 100mS anyway.

It could just be the case that the dialler is sending tones that are too
short either for the remote end decoder or where there is some form of
audio a.g.c. which cannot track the tone amplitude variations.

I suspect that if Brian can change the tone duration to 100mS it might
just work.

[David Higton]

In message <sk4803$6m5$1...@dont-email.me>

Woody <harro...@ntlworld.com> wrote:

>
> There is one more point on which Brian can probably answer. DTMF diallers
> on the mouthpiece as we are discussing, if using a programmed (i.e.
> memory) number, often dial the full number sequence using their own
> timing. DTMF is specified to a minimum per tone time of 40mS, but some
> decoders can even require as much as a 100mS tone to get certain decoding.
> Over many years of radio to telephone interfacing (which is not so
> different) I found that in most cases 70mS will work reliably but I
> usually set it to 100mS anyway.
>
> It could just be the case that the dialler is sending tones that are too
> short either for the remote end decoder or where there is some form of
> audio a.g.c. which cannot track the tone amplitude variations.
>
> I suspect that if Brian can change the tone duration to 100mS it might
> just work.

Yes, been there (or a very similar place). I agree with what you say.

David

[Andy Burns]

Brian Gaff wrote:

> Exactly my point.

I think the last time I used a DTMF dialling device, it was a Psion 3a

[Woody]

For the record the difference is quite noticeable audibly. 40mS tones
(as used in EEA radio selective calling) just sound like a fast sequence
of pips, whereas 100mS tones (as used in CCIR signalling - mostly
marine) sounds positively musical.
For comparison the first five of the six GMT pips are 100mS each.

[Richmond]

Maybe you could use a smart phone with a voip app, and then get the
smartphone's voice recognition to dial for you.

[Brian Gaff (Sofa)]

Yes another case of not thinking it through and probably inconveniencing the
disabled due to nobody actually looking at the usage of such devices. The
beauty of the Parat device, for example is that if you were at a mates house
any landline could be used to dial from your portable directory without
having to remember groups of numbers as they are dialled. Ideal for those
with poor eyesight and or dementia.

Brian

--

This newsgroup posting comes to you directly from...
The Sofa of Brian Gaff...
bri...@blueyonder.co.uk
Blind user, so no pictures please
Note this Signature is meaningless.!

"David Higton" <da...@davehigton.me.uk> wrote in message
news:4444437a59.DaveMeUK@BeagleBoard-xM...

[Brian Gaff (Sofa)]

No it did not but the agc effect in the exchange and the lack of one in voip
could make the tones so distorted that they cannot be read.
I got rid of Voip here, and while Virgin are still happy to let me have
analogue fine but it does mean that on some phones in other peoples houses
the effect is more unpredictable. There is certainly a volume adjust in the
dialer, but the speed and duration are mostly similar to a telephone, I
think.

Brian

--

This newsgroup posting comes to you directly from...
The Sofa of Brian Gaff...
bri...@blueyonder.co.uk
Blind user, so no pictures please
Note this Signature is meaningless.!

"Woody" <harro...@ntlworld.com> wrote in message
news:sk4803$6m5$1...@dont-email.me...

[Brian Gaff (Sofa)]

That sounds very convoluted, if you are going to do that then you would be
pointless having the landline altogether and I suspect its not going to work
any better as it will be using its own tones not yours.
I will have to find a guinea pig with a modern voip from bt or virgin and
test it.

Brian

--

This newsgroup posting comes to you directly from...
The Sofa of Brian Gaff...
bri...@blueyonder.co.uk
Blind user, so no pictures please
Note this Signature is meaningless.!

"Richmond" <rich...@criptext.com> wrote in message
news:84mtnez...@example.com...

[notya...@gmail.com]

Just tested this by ringing my POTS line using VoIP from my mobile. Once connected DTMF tones were transmitted. Same from my Gigaset DECT phones too.

[Richmond]

"Brian Gaff (Sofa)" <bri...@blueyonder.co.uk> writes:

> That sounds very convoluted, if you are going to do that then you would be
> pointless having the landline altogether and I suspect its not going to work
> any better as it will be using its own tones not yours.
> I will have to find a guinea pig with a modern voip from bt or virgin and
> test it.
> Brian

I don't understand that response. I thought the purpose was to make a
phone call by speaking rather than dialling a number. If you use a smart
phone and voip then you can do that and it doesn't matter what the tones
are. It's not the same as making a mobile call because it will be using
wifi and broadband.

The purpose of the landline will be to carry the broadband and voip, as
there won't be anything else after conversion to voip anyway.

[Brian Gaff (Sofa)]

No I cannot understand why everyone is being thick on here. Sorry, maybe its
the way I tell em as someone used to say.
Parrat Voicemate allows you to put all your numbers into it and call them
by repeating the name. The dialling happens over the mouthpiece of a
standard land line phone with no issues. Then you connect that same phone to
the voip system that bt and virgin are going to move over to, and the tone
recognition is very patchy if it works at all.For us blind folk who cannot
read the land line phones lists on the display its a very useful way to
always be able to call if you do not have a mobile phone, which as we all
know is a great learning curve for a blind user.
If this is what will continue to occur then it seems that you might as well
ditch the land line n and get a non smart mobile and put all the names into
it. You have a limited choice, however since very few can recall numbers by
voice or indeed speak at all. Thus, you have to get a smart phone put on
talkback or voice over and learn the blind gestures and get a bluetooth
keyboard to enter in all your names and numbers as the on screen keyboard is
a pain for a blind user.

Brian

--

This newsgroup posting comes to you directly from...
The Sofa of Brian Gaff...
bri...@blueyonder.co.uk
Blind user, so no pictures please
Note this Signature is meaningless.!
"Richmond" <rich...@criptext.com> wrote in message

news:844k9im...@example.com...

[David Higton]

In message <ske2ba$jm$1...@dont-email.me>

"Brian Gaff \(Sofa\)" <bri...@blueyonder.co.uk> wrote:

> No I cannot understand why everyone is being thick on here. Sorry, maybe
> its the way I tell em as someone used to say.
>
> Parrat Voicemate allows you to put all your numbers into it and call them
> by repeating the name. The dialling happens over the mouthpiece of a
> standard land line phone with no issues. Then you connect that same phone
> to the voip system that bt and virgin are going to move over to, and the
> tone recognition is very patchy if it works at all.For us blind folk who
> cannot read the land line phones lists on the display its a very useful
> way to always be able to call if you do not have a mobile phone, which as
> we all know is a great learning curve for a blind user.
>
> If this is what will continue to occur then it seems that you might as
> well ditch the land line n and get a non smart mobile and put all the
> names into it. You have a limited choice, however since very few can
> recall numbers by voice or indeed speak at all. Thus, you have to get a
> smart phone put on talkback or voice over and learn the blind gestures and
> get a bluetooth keyboard to enter in all your names and numbers as the on
> screen keyboard is a pain for a blind user.

I can understand the challenge you face. Parrot Voicemate would appear
at first to be a good solution, but, if the company is the Parrot that
I've been looking up, it's no longer supported, so we can't find any
more information that might help get this working better for you. (If
the product is still supported, I'm happy to be corrected.)

There are several possible contributors to poor reliability:

1) The acoustic coupler, which seems an inevitable part of the signal path.

2) We don't know anything about the timing of the DTMF, nor can we try
altering it experimentally.

3) If you're using the Voicemate with a VoIP or DECT phone, the codec in
use between the handset and the adapter is very likely to degrade the
ability to decode the DTMF. That would not be so in the traditional
case of a POTS line.

The above is a pretty bald statement, unfortunately with not much in the
way of solutions in sight.

Is there a realistic alternative to Voicemate?

David

[Theo]

David Higton <da...@davehigton.me.uk> wrote:
> Is there a realistic alternative to Voicemate?

Something the might be similar is a Bluetooth dialler. You enter your
numbers into it, and pair it with your cordless / mobile / tablet / PC.
Then, when activated, it dials the number over Bluetooth.

While they exist, I didn't manage to find a voice activated one at a quick
glance. But maybe there are some out there. At least they aren't dependent
on DTMF.

However I wonder whether it might be simpler to have an Android phone with
an integrated keyboard. There are some from Blackberry and others, for
example the Psion 5 derived ones like the Cosmo Communicator. While I'm
sure you can navigate Android with voice prompts alone, I'd have thought a
physical keyboard would be much more pleasant.

There are also phones with physical dialpads - they might suffice for calls
only but a fully keyboard might be better for things like messaging.

Theo

[Woody]

I doubt the audio path of codec would have much if any effect on the
audio given the highest frequency is 1477Hz - still well into speech band.
My money still goes on tone duration which Brian has not yet commented on.

[Richmond]

"Brian Gaff (Sofa)" <bri...@blueyonder.co.uk> writes:

> No I cannot understand why everyone is being thick on here.

Well fuck off then.

[David Higton]

In message <84h7dgz...@example.com>

Oh dear.

Some respondents appear not to have noticed that the OP is blind, which
renders a smartphone nigh on useless to him.

David

[David Higton]

In message <skerip$p7$1...@dont-email.me>

The frequencies are all well within range - that's not the problem.

With a lossy compressing codec chain, the frequency that comes out is
not necessarily the frequency that went in. Not a problem with G.711,
but can be with some of the others.

David

[David Woolley]

On 16/10/2021 21:33, David Higton wrote:
> With a lossy compressing codec chain, the frequency that comes out is
> not necessarily the frequency that went in.

In particular, they won't handle two non-harmonically related tones
well, as that is not something the human vocal tract can produce.

[Richmond]

I've seen a blind person using a smart phone.

Being blind is no excuse for calling people thick.