Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Voice Modems

computer.rip

Post reply on HN
Print view View original

11–18 of 18 posts

Re: Voice Modems

#11

Post by jcims »

pm215 wrote: post #3

For a long time in the dialup era my "answering machine" was a US Robotics voice-capable modem attached to my home Linux PC, with some scripts to make it pick up after N rings, play a message, record whatever the caller said, and then email me the resulting sound file. The Linux support for it included DTMF tone recognition, so I added in a quick hack so that if I sent it the right pin code during the "please leave a…

I was going to use that to start a private voicemail company in my little rural town. Had the name and everything ready to go (mailvox!) but I was too broke to afford the second phone line xD.

Plus in retrospect I'm sure it would have been used almost exclusively for illicit purposes. But that wasn't really something I had thought of back then.

Re: Voice Modems

#12

Post by rasz »

marlburrow wrote: post #8

The DLE-escaping problem the article mentions — where 0x10 bytes in PCM audio had to be doubled to 0x10 0x10 — is the same class of bug that plagued the Hayes +++ escape sequence in data mode, and it is striking that it was never properly solved. The fundamental issue is that the Smartmodem architecture multiplexed control and bearer on a single serial channel, and every framing scheme for doing that with an 8-bit-cl…

Didnt later modem standards introduce HDLC like mechanisms? MNP and later LAPM packetize modem traffic thus you can control what is raw data and what is control layer.

Re: Voice Modems

#13

Post by bigfatkitten »

rasz wrote: post #12
marlburrow wrote: post #8

The DLE-escaping problem the article mentions — where 0x10 bytes in PCM audio had to be doubled to 0x10 0x10 — is the same class of bug that plagued the Hayes +++ escape sequence in data mode, and it is striking that it was never properly solved. The fundamental issue is that the Smartmodem architecture multiplexed control and bearer on a single serial channel, and every framing scheme for doing that with an 8-bit-cl…

Didnt later modem standards introduce HDLC like mechanisms? MNP and later LAPM packetize modem traffic thus you can control what is raw data and what is control layer.

Not in any widespread manner. Legacy compatibility won as far as the interface between the modem and the computer was concerned.

Re: Voice Modems

#14

Post by ssl-3 »

glasss wrote: post #7
EvanAnderson wrote: post #4

Earlier quoted context omitted.

I wrote some stuff in Turbo Pascal for DOS to do something like this (albeit it didn't email files-- it just dumped them into a directory on the disk). My parents had two phone lines so making test calls from a real phone was easy. I just had to go around the house and turn off the ringers on all the phones so I wouldn't wake anybody doing test calls in the wee morning hours. I didn't understand the sample format so…

A voice based BBS could probably trace some kind of cultural or technical lineage to TikTok today, interesting to think about.

There were some things that came kind-of close.

TellMe was one: Call the number, ask it questions, get answers. Part of my normal commute for a time involved calling TellMe to get the weather for the day on a Nokia dumb phone once I got settled into the drive.

Goog411 was another one. You could call Google, ask it questions, and get summaries of search results along with answers for a distinct questions (a lot like LLMs do today). I distinctly recall standing in the supermarket looking at large and inexpensive hunk of meat that was labelled as a "Boston Butt Roast", and calling Goog411 to find some common uses for it. (It did give me confidence to buy it, and we did cook it and eat it. It was lovely.)

These things worked well for that brief moment in time when cellular calling minutes were either plentiful or unlimited, when smartphones didn't commonly exist, and when mobile data was ludicrously expensive.

Re: Voice Modems

#15

Post by gwbas1c »

marlburrow wrote: post #8

The DLE-escaping problem the article mentions — where 0x10 bytes in PCM audio had to be doubled to 0x10 0x10 — is the same class of bug that plagued the Hayes +++ escape sequence in data mode, and it is striking that it was never properly solved. The fundamental issue is that the Smartmodem architecture multiplexed control and bearer on a single serial channel, and every framing scheme for doing that with an 8-bit-cl…

> is the same class of bug that plagued the Hayes +++ escape sequence in data mode, and it is striking that it was never properly solved.

When I ran a BBS, it was critical to configure the modem and software to use signaling lines (I forget the term's name, something like the rx or tx line) to signal commands.

I'm surprised that voice modems weren't using these, as they were common with external modems.

Re: Voice Modems

#16

Post by cbluth »

ssl-3 wrote: post #14
glasss wrote: post #7

Earlier quoted context omitted.

A voice based BBS could probably trace some kind of cultural or technical lineage to TikTok today, interesting to think about.

There were some things that came kind-of close. TellMe was one: Call the number, ask it questions, get answers. Part of my normal commute for a time involved calling TellMe to get the weather for the day on a Nokia dumb phone once I got settled into the drive. Goog411 was another one. You could call Google, ask it questions, and get summaries of search results along with answers for a distinct questions (a lot like L…

Goog411 had sms and travel directions, as tap into something like google maps. I used it to travel several interstate trips back in the day

Re: Voice Modems

#17

Post by stevetron »

There is at least on class of voice modems the article didn't mention: The type 407 modem. In the 1980-1984 time period, I was part of a project to build a system for approving-disproving checks at some point of sale. The customer came to us, and wanted a way for a retailer to use a touch-tone phone to input a check routing code, have their database provide a determination of the check being good or bad, and returning a message via voice sysnthesizer to the customer on the telephone. In a nutshell, anyway. Their competitors kept requiring the retailers to purchase phones with integral card readers (for reading mag stripe credit cards).

I designed and built the box with the speech synthesizer, and it sat between the customer's computer, and a type-407 modem.

The Bell 407 modem was a large rack of many modems. Too pricey. Too big a hardware investment if you just wanted one incoming line. But we found something called a Tuck 407. The Tuck 407 utilized a 300-baud modem chip, and the one we had only could transmit in this mode. We didn't use this mode. We used the mode that utilized a touch-tone decoder and fed the digits out the modem's serial port. The modem also featured an input for audio, which it fed back into the phone lines. Thus, a voice modem. But the price of teh Tuck 407 was still somewhat pricey at $400 each. I only ever saw two of them cross my workbench, and they were resold to the customer.

My company built their own type 407 modem. They wanted basically a pc board level product that had the same footprint as the voice synthesizer board, so they could both be placed in the same enclosure (card rack). Eventually, that was done. It also chewed up about $30K in R&D costs. But it worked very well. It would have implemented the AT-style instruction set, but nobody was sure how to shoe-horn that on top of the hardware we'd built, that relied on extra control pins of the RS-232 connector into the modem. Several sample units were built, the product passed it's FCC part 15 and part 68 requirements, and only 2 initial sample modem boards ever made it to the customer that had also bought the two Tuck 407 modems.

The plans for the 407 modem got burned up in a fire. Almost like in a novel, it was so classic.

Re: Voice Modems

#18

Post by dfe »

gwbas1c wrote: post #15
marlburrow wrote: post #8

The DLE-escaping problem the article mentions — where 0x10 bytes in PCM audio had to be doubled to 0x10 0x10 — is the same class of bug that plagued the Hayes +++ escape sequence in data mode, and it is striking that it was never properly solved. The fundamental issue is that the Smartmodem architecture multiplexed control and bearer on a single serial channel, and every framing scheme for doing that with an 8-bit-cl…

> is the same class of bug that plagued the Hayes +++ escape sequence in data mode, and it is striking that it was never properly solved. When I ran a BBS, it was critical to configure the modem and software to use signaling lines (I forget the term's name, something like the rx or tx line) to signal commands. I'm surprised that voice modems weren't using these, as they were common with external modems.

It was drop DTR. Most current documentation tells you that dropping DTR means the modem should hang up. That was an option, maybe even the default. But you could AT&D1 to make dropping DTR return you from data mode to command mode. It's so important there's even a DIP switch on the USR Courier to enable this mode.

https://support.usr.com/support/3453b/3453b-crg/chap%208-con...

I almost can't believe the USR site is still up. This is something I remember doing to run a BBS over 30 years ago. I still have my Courier with the brass not for resale plate.

Post reply on HN