OT: the current state of telephone hybrids

[Dave Labrecque]

That's the likely culprit. It works, but it's not necessarily consistent. Usually, all it takes to fool the hybrid is for the company's gear to insert the ring-tip voltage on the "POTS" line, because that's all that 99% of end-user equipment needs.

Sidebar: I'd always thought of the ring conductor on a TRS connector to be named that because of its shape (much like "tip" and "sleeve"). I have an engineer friend who told me that the name actually comes from the voltage applied to that conductor used to make the phone ring. The "ring voltage."

Apocryphal?

[CurtZHP]

Sidebar: I'd always thought of the ring conductor on a TRS connector to be named that because of its shape (much like "tip" and "sleeve"). I have an engineer friend who told me that the name actually comes from the voltage applied to that conductor used to make the phone ring. The "ring voltage."

Apocryphal?

Could very well have some truth to it. Some of the technology we use in audio originally came from telephony. The 600-ohm impedance standard is a hold-over from early telephone hardware. The TT-type (Tiny Telephone) patch bay connectors are just miniature versions of the original telephone patch cords. (This is also why 1/4" plugs and jacks are sometimes referred to as "phone" connectors...) The concept of audio equalization also got its start in the telephone industry.

Here's some interesting reading....

http://www.epanorama.net/documents/t...interface.html

[jmh]

I am not exactly sure what your objectives are, but I am going to chime in. If your facility has a modern call manager, or PBX, it may have the facilities to do what you want. I (in my - uck - day job, I am a full time phone system administrator at a school district). We run asterisk and there are many things that you can do that might be useful. For example, you can run mix-monitor which can dump one or the other half of a conversation (or both) into a file, or you could always set up calls to conference to a sound card or digital stream, a sip device, whatever. There are all kind of possibilities and almost all of them would give about the best fidelity your going to get compared with another device (at least with the half of the conversation that is not local) since whatever format the phone system is passing is available for your use - typically 8000 sample rate compressed into an 8 bit ulaw stream. But asterisk can also switch high sample opus (among many formats) streams which can have quite high fidelity too - not that you currently get endpoints calling in supporting it - but with an IP opus stream experiencing little delays, you theoretically can already get supreme quality.

At my school, I have PRI circuits which deliver the signal in digital format so the techniques I mentioned above would yield optimum results. In a pots scenario, we are talking about an analog signal - but as the audio has traversed the telephone network, it has been digitized and compressed in some codec (and often several different codecs and conversions may have taken place).

Anyway, if you have someone who administers your phone system, it is worth having a conversation as you already have a potential solution. That said, you invest in programming and set-up time instead of a device.

John

...a little footnote about telecommunications audio codecs. Codecs utilized in phone systems have special requirements - and IP traffic adds additional issues - a couple of them are:

Low latency conversion algorithms (they can't study too many samples before doing their thing)
They have to be able to recover from lost data without causing to many artifacts. If you loose an IP packet, you can't wait for a retransmission, you have to wing it and move on.
...and a bunch of other considerations.

[Dave Labrecque]

Could very well have some truth to it. Some of the technology we use in audio originally came from telephony. The 600-ohm impedance standard is a hold-over from early telephone hardware. The TT-type (Tiny Telephone) patch bay connectors are just miniature versions of the original telephone patch cords. (This is also why 1/4" plugs and jacks are sometimes referred to as "phone" connectors...) The concept of audio equalization also got its start in the telephone industry.

Here's some interesting reading....

http://www.epanorama.net/documents/t...interface.html

Let's not forget punch blocks. I still use 'em. Although with outboard gear becoming less common, I'd think these would be fading in usage.

[Dave Labrecque]

I am not exactly sure what your objectives are, but I am going to chime in. If your facility has a modern call manager, or PBX, it may have the facilities to do what you want. I (in my - uck - day job, I am a full time phone system administrator at a school district). We run asterisk and there are many things that you can do that might be useful. For example, you can run mix-monitor which can dump one or the other half of a conversation (or both) into a file, or you could always set up calls to conference to a sound card or digital stream, a sip device, whatever. There are all kind of possibilities and almost all of them would give about the best fidelity your going to get compared with another device (at least with the half of the conversation that is not local) since whatever format the phone system is passing is available for your use - typically 8000 sample rate compressed into an 8 bit ulaw stream. But asterisk can also switch high sample opus (among many formats) streams which can have quite high fidelity too - not that you currently get endpoints calling in supporting it - but with an IP opus stream experiencing little delays, you theoretically can already get supreme quality.

At my school, I have PRI circuits which deliver the signal in digital format so the techniques I mentioned above would yield optimum results. In a pots scenario, we are talking about an analog signal - but as the audio has traversed the telephone network, it has been digitized and compressed in some codec (and often several different codecs and conversions may have taken place).

Anyway, if you have someone who administers your phone system, it is worth having a conversation as you already have a potential solution. That said, you invest in programming and set-up time instead of a device.

John

...a little footnote about telecommunications audio codecs. Codecs utilized in phone systems have special requirements - and IP traffic adds additional issues - a couple of them are:

Low latency conversion algorithms (they can't study too many samples before doing their thing)
They have to be able to recover from lost data without causing to many artifacts. If you loose an IP packet, you can't wait for a retransmission, you have to wing it and move on.
...and a bunch of other considerations.

All very impressive -- and over my head. But thanks.

My client's situation is very simple: she has a podcast recording studio in her home with an IP phone coming off a router. No call manager. Looks like we're moving to a POTS line because of things we don't like about the IP phone setup. Either way, I just was asking what the hybrid options are these days. We need to null out the outgoing audio from the incoming audio (there's outgoing audio mixed in there so that we can hear ourselves in the earpiece along with the caller, as I understand it), and that's what a telephone hybrid does. It's not a simple task, but digital technology has made things a lot better in recent decades.

[jmh]

All very impressive -- and over my head. But thanks.

We need to null out the outgoing audio from the incoming audio

That is going to be harder with a pots line as it is bi-directional on one pair.

If you felt like doing some expermentation, you could try tapping the IP phone's handset speaker and pass the mic signal through a mixer (or the headset port which most of these devices have). No guarantees, but if it works, that should give you two isolated signals - which would not be the case if you tried it on an analog phone...

[Dave Labrecque]

That is going to be harder with a pots line as it is bi-directional on one pair.

Right. That's why the telephone hybrid was invented. It's quite a feat to pull off, technically, I think. Broadcasters have been using them for decades.


https://en.wikipedia.org/wiki/Telephone_hybrid


[/QUOTE]If you felt like doing some expermentation, you could try tapping the IP phone's handset speaker and pass the mic signal through a mixer (or the headset port which most of these devices have). No guarantees, but if it works, that should give you two isolated signals - which would not be the case if you tried it on an analog phone...[/QUOTE]

That's exactly the way we've been doing it for years (taking the caller audio from the headset jack), but the host signal is only about 10 dB below the caller's, there. Not great. Certainly not "isoloated" in any strict sense. But you just gave me an idea (after I first misread what you wrote): tapping the base's speakerphone audio has never occurred to me. You'd think the phone's designers would be trying really hard to keep the host audio out of there! Hmmmm...

Of course, a nice digital hybrid is going to do spectral balancing and stuff, too (for call-to-call consistency), so that's probably still the best route.

I'd be interested to hear, though, if anyone tried a tap off of the speakerphone's signal (which is not a line-level thing -- how would that work?).

[jmh]

The speakerphone circuit will require a bunch of signal processing that the handset or headset will not (or at least not as much). Gate, expander or ducker might be part of it - so I would suspect that would be the most problematic port to work with...

Another thing to be aware of is echo cancellation (where an endpoint is responsible for canceling echo generated on it's end) has long been a part of telephone circuitry. It has gotten a whole lot weirder since IP came into the picture as the longer and or variable delays which IP networks introduce can make what would be an unnoticeable echo on a pure analog conversation unbearable...

[jmh]

That's exactly the way we've been doing it for years (taking the caller audio from the headset jack), but the host signal is only about 10 dB below the caller's, there.

Are you catching the handset mic and speaker on separate channels? If the handset is attached to an IP phone, you should get 2 pretty distinct signals...

(as opposed to an analog device where they will be smeared together - I'll have to look up the hybrid phone too, as I don't use them, I am not sure what they do)

[jmh]

I took a look at the hybrid link. Yea, one of the problems that phone users encounter nowadays is related to what I had mentioned about echo cancellation being applied to one half of the conversation - that is your end is responsible for getting rid of echo propagated on your end - and the remote endpoint for their end. But it is like the wild west now where you don't know what the **** is happening on the far end - and I would expect this to create issues with a hybrid device (as first described in the wikipedia article).

It is analogous to a situation I encounter with a fax server (a technology developed before any significant delays existed in telephony) I manage which handles many many faxes a day and performance gets more and more erratic as more people get their service from non-traditional phone providers.