-- Eugen* Leitl <a href="http://leitl.org";>leitl</a> ______________________________________________________________ ICBMTO: N48 04'14.8'' E11 36'41.2'' http://www.leitl.org 57F9CFD3: ED90 0433 EB74 E4A9 537F CFF5 86E7 629B 57F9 CFD3 ---------- Forwarded message ---------- Date: Fri, 1 Feb 2002 14:52:16 -0800 (PST) From: Jeffrey Streifling <[EMAIL PROTECTED]> To: Alejandro N�stor Vargas <[EMAIL PROTECTED]> Cc: [EMAIL PROTECTED], [EMAIL PROTECTED] Subject: Re: Speak-Freely and a Telephone > > I read your message with great interest because I wanted to connect > > Speak-Freely with a telephone also. I do know there were a group of > > developers creating software drivers for this purpose, but for some > > reason the development has been stopped. > > Well... I hav not much time but I can help if the project is already > started. If you can help, may be we could continue the project. Where do > you seen this? > > > With this type of setup, Speak-Freely becomes a very very powerful > > communications tool. I have done this. I set up a Speak Freely to POTS gateway that allowed somebody to connect to the machine over the Internet and make an outbound call from the machine, which ran unattended. This turns out to be a fairly difficult thing to do. (1) Because there is no way to manage line turnaround from the remote telephone, you must do everything the full-duplex way. (2) To run unattended in an obvious way, you will probably want to base your system on a Unix-like platform, BUT a lot of Unix-like platforms restrict to you half-duplex. (3) You cannot wire a sound card to the phone line in the obvious way and expect to do full duplex. Everything you drive onto the wire from the DAC will feed back into the ADC real loud, making communication impossible. To solve this, you will need a specially wound transformer called a "hybrid coupler". Internally, they are not all that complex, and they show up in a variety of telephony equipment, but the kind of thing you will want for this job is a bit of a rare item. The best way to do this is to talk to find your friendly amateur radio operator and ask how to get a phone patch. You will still need to know how to wire up small amplifiers and resistor networks to handle the impedance transformations. (4) If you are connecting to a POTS line, you need a way to control your output impedance. High impedance = on hook; low impedance = off hook. If your hybrid is high impedance, you can put a Hayes style modem in parallel with it to manage dialing and hookswitching. If your hybrid is low-impedance (forcing the phone off-hook), you will need to retrofit it with a relay to hang up the line with, and make the necessary arrangements for controlling it. (Remember how pulse dialing works?) (5) There are several cans of worms on the computer side, including management, security, CPU management, and others. I never did get a good interface worked out for the whole mess. Rather than pull your hair out, you should consider getting hardware that is suited to the job (Quicknet makes something called the Linejack, and there is a company called Voicetronix which would be useful for larger setups). Rather than use Speak Freely, which is oriented to interactive use, try something along the lines of the tools from www.openh323.org. H.323 does not really address encryption (to my knowledge); use CIPE. Actually, the encryption in Speak Freely (at least the current Unix version) has a number of problems. The two grossest problems are the fact that the one-time pad is not one time (it's one time per packet) and the fact that the IDEA encryption uses the cipher feedback mode with an all zero initialization vector, thereby encrypting the first eight bytes by XORing them with a constant (the not-so-onetime-pad problem, round two). This trivially "gives away the farm". More minor issues include the fact that text chat is not encrypted (from what I can tell), and DES has too short a key to be of much use anymore. The moral of the story is, "Use Blowfish!". (Is there a fix in the works?) Anyway, my project never did work all that well -- it was short on CPU power, my prototyped (unshielded) circuits picked up a lot of noise, the interface was clunky, grounding was problematic, and keeping the signal amplitude at reasonable levels through the whole apparatus turned out to be a nightmare. Good luck; you'll need it! Jeffrey Streifling <[EMAIL PROTECTED]> * * * To unsubscribe from this mailing list, send E-mail containing the word "unsubscribe" in the message body (*not* as the Subject) to [EMAIL PROTECTED]
