Industry-Standard VOIP Phone Using All Free Software

Ralf Ackermann writes: "Voice over IP on a HardPhone running Linux and just using Open Source software became real. We have sucessfully installed and tested (interoperability with Cisco 7960 as well as Pingtel xPressa in an environment with a partysip SIP registrar and proxy) the linphone SIP phone on a StrongARM based TuxScreen. Here is the link describing the steps for others to use the setup as well: TuxScreen running SIP. All the infos for setting up a comparable installation can be found on the URL, please also feel free to ask or drop opinions. Many thanks to the linphone developers as well as to my student Florian Winterstein (for working on a console linphonec version). The setup (on a StrongARM system) is well suited for PDA (iPAQ) or wearable environments as well."

Comments

What about SNOM?

[DragonWyatt]

Check this out for another linux-based VoIP, standards-compliant (both SIP *and* H323) phone.

It's been out for quite awhile, over a year. My company is a reseller. They're cheap (~ $199 each) and they rock.

One step closer...

[Justen]

The company I work for has been looking at VoIP for several months now. Since we have several offices across the country, it would be very advantageous, technologically, for us to use VoIP for the end-user phones, rather than the hodge-podge of systems we use today.

Unfortunately, the prohibiting factor has been the cost of the phones themselves. The cost for an actual system is within reason, but some VoIP telephones run into the $700 range.

At this pricepoint, it seems much more affordable and reasonable. And while the GUI would need work to make it dummy-friendly, we have no shortage of graphics designers and programmers who could make that work.

One step closer to VoIP from beginning to end makes me happy. And I know it'd make our CFO happy, too. =)

jrbd

Why so low market penetration?

[peterdaly]

I do not understand why IP phones
a) have so low market penetration
b) cost so much
-- I know part of the problem with a is b.

What I think really is needed is a low cost, high quality server system for one of these systems. Based on what little research I have done, it costs almost as much, if not more, for hardware for a small office system as it would to get a real small PBX like phone system.

I don't think a phone really needs a 9" LCD screen, or whatever was in the screen shot, but the "Java Phone" from the other company has a screen size perfect for the company phone directory. That I think could be the "killer app" for these things.

Anyway, anyone know of low cost PBX software (if that is even what is is called in the IP phone world)? Open Source, under a BSD like license would be cool, and lower the barrier to market entry for companies wanting to roll a system like this out. Of course, cards to hook up to a POTS connection would also be needed. Voicemail over the web, via shoutcast or something would rock. I havn't listened to my voicemail at work in 3 months. With a better interface, I may stop refusing to use it.

-Pete

The next step

[FueledByRamen]

This is fine for intraoffice communication, but what about the real world? I think that a solution for connection VoIP -> POTS could be easy, with a little knowledge of Linux drivers.

Get an older box (P2 400 or so), with plenty of PCI slots, and preferrably an onboard NIC also. Get some Winmodems equal to the # of pci slots.

WinModems, even in all of their Microsoft-sponsored godless evil towards open source platforms, are basically A/D and D/A converters hooked to a phone jack. It should be relatively simple to talk (no pun intended) to them in software and use one as an interface to POTS. It has all of the neccesary hardware, and writing a sound driver for it shouldn't be too difficult. A brand of WinModem with fairly standard hardware could be decided on by the implementer, and drivers written for that. (Winmodems? Standard? ...)

Client software with available source code could be modified to use those, as well as control the phone-line functions. Just run an instance per WinModem.

Honestly, I think that this could work, and it would be a great hack to accomplish. Anyone fancy a go at it?

Re:The next step

[elandal]

I think that'd be pretty inefficient. Rather, get a card that talks to PRI. Don't know of any such cards, but they're very likely to exist. And if they don't exist, find a small engineering shop nearby and ask what would it take to put one together.

Think about it.. A box with six PCI slots and mobo-NIC using six WinModems could handle six phone lines - old analog lines at that. While a single PRI-card could handle 30 lines, and considering that each ISDN line is just 64000bps (8000 samples at 8bits per second - if I remember correctly), it doesn't really all that much computing power, so a single PC should easily handle it.

Of course a small shop doesn't need 30 phone lines, and could do with the WinModem-based setup, and those that need can get the VoIP from the telco, not needing their own VoIP-POTS conversion.

VOIP for PCs?

[fopa]

Any info on VOIP for a regular PC? I need international long-distance access. I saw this Ask Slashdot article that recommended Speakfreely, but I haven't had time to try it.

Any advice?

FYI

[Cyberdeck]

Just a couple of points I'd like to comment on.
Open Source VoIP and telephony is tracked at http://www.linuxtelephony.org.

VoIP hardware (PCI cards) is IIRC $79, gateways are $179, drivers have been in the kernel since (again IIRC) 2.2.16. http://www.quicknet.net.

VoIP in the U.S. is almost pointless because the PSTN is too good. No one wants near perfect when perfect is cheap and easy. In the third world, if you can get a phone circuit it averages $1.27 per minute, whereas VoIP through a hop-off provider like Net2Phone (http://www.net2phone.com) runs average about $0.23 per minute or less. Straight IP to IP (like across a VPN from company branch to branch) is just the cost of the ISP (usually flat rate). So VoIP cafes are a popular way for the non-super-rich and powerful to make calls to their relatives in the first world.

Personally, the VoIP calls I have made have an almost imperceptible latency problem and sounded *far* better than any GSM call I have ever heard. Then again, these calls were during business hours so net congestion was not an issue for me.

The Ogg Vorbis has a low bit rate mode that is useful for VoIP telephony, and is grossly better than GSM to my ear.

Finally, VoIP is used by the big players here in the US. Qwest and Sprint use VoIP in preference to ATM due to cost of the equipment (Bits per second/price of hardware. IP is more efficient than ATM due to less overhead). Any cross-country call is VoIP nowadays.

Just my two bits.

-C

Re:FYI

[Bookwyrm]

Strictly speaking from a bandwidth usage point of view, IP is not more efficient than ATM for voice, unless you allow for higher latency per chunk of voice data.

That is, ATM has a fixed, roughly 10% header overhead per cell compared to data payload. IP can have lower relative overhead by having larger packets verses the IP header size -- *however* the time it takes to gather the data to fill one large IP packet increases the delay.

For instance, if you are using an 8:1 compression codec, then that 64K bits-per-second voice becomes 8Kbps, or 1KBps. You can fit that entire one 1K bytes into a single IP packet, yes -- but you have to wait one entire second (the time it takes to gather that second of voice) before you can send the packet. With 8:1 compression, each byte of data is roughly 1 millisecond of voice -- each byte you pause to gather adds one millisecond delay before you send the data.

ATM becomes more efficient for voice as the payload size decreases per IP packet below around 300 bytes (i.e. 300 milliseconds delay in voice transmission between the first byte encoded and the last byte encoded before the packet is sent.) IPv6 will have an even poorer efficiency, of course, due to the larger header.

This is not comparing the cost of ATM network hardware to IP networks, just commenting that from a strict data point of view (i.e. bits per second of payload), ATM is going to be have better efficiency/quality for voice data. Now, for mixed voice and data networks, that may not be the primary concern.