Mozer (Digitalker) chip MM54104 speech data encoding

Greetings:

I am hoping to generate speech data for the old NS MM45104 (Mozer) Digitalker chip; sources or binaries for an encoder or a complete description of the process would help the most.

A description of the technology (broadly) can be found in this patent:

formatting link

which applies to a similar but somewhat different implementation from that used in the MM45104.

This site discusses the device implementing the technology described in the patent:

formatting link

And here is that site's author's decoder program;

formatting link

When time permits, I will try decoding the data from some speech ROMs that I have saved from an old product to see how close they come to the format described above. In any case encoding is much more difficult and the described format is known to be different from that used in the Digitalker chip.

Regards,

Michael

Reply to
msg
Loading thread data ...

I find that hard to believe, since your headers show:

User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.7.2) Gecko/20040804 Netscape/7.2 (ax)

Are you running on a Transmeta chip?

Reply to
larwe

s/45104/54104/g

Sorry for the typo; it's -20 deg. F. here and barely above that indoors.

Regards,

Michael

Reply to
msg

No, but that would indeed be ironic. I am typing at a very cold laptop which is a terminal services client to a host in a warm place...

Actually the host is in the 'dome' -- a converted Ground Control Approach radar radome which houses the local infrastructure and is the only spot on premises above +40 deg. F. at the moment (mostly from machine-generated heat). When I just can't endure the cold any longer I retreat to its interior to work amidst the din of fans, et al. Everywhere else suffers from lack of heat at the moment (except for a 500W heater at my feet) since the wood I expected to burn is too green and consumes more calories than it emits. Feel free to do a GIS lookup on the IP of this post and check the temperature :-). Fortunately I have no large HDDs running in the really cold places; I've had serious problems when the impregnated bearings aren't mil-spec temp rated. Ever try to power on an ST-506 cooled below freezing (it screams like a banshee until it succumbs).

Regards,

Michael

Reply to
msg

I fail to see the connection with his local temperature. I can remember -40 outside Montreal and in Deep/Chalk River. Not in the least unusual.

--
Chuck F (cbfalconer at maineline dot net)
   Available for consulting/temporary embedded and systems.
Reply to
CBFalconer

Jeff Frohwein has done that for the SSR1 and SHRL ROM:

formatting link
It has probably not been done for the SSR2.

Besides Telesensory and National Semiconductor there is a third chip: the "talking clock" UAA1003 from ITT/Germany. Easy to use, as the ROM is on the chip. There were German, French, UK versions. Distribution limited to Europe due to Mozers US-patents ?

4214125 filed 1977 issued 1980 is certainly the basis. But for the Digitalker other Mozer patents may be relevant too: 4384170 filed 1979 issued 1983 4384169 " 4433434 filed 1981 issued 1984 4435831 "

In books more readable then the patents :

Bristow "Electronic Speech Synthesis" p 137 - p 147 Sickert "Automatische Spracheingabe und Sprachausgabe" M&T 1983 p 169 - p 175 ( obviously in german ) Can do a scan if necessary.

National did some ApplicationNotes: AN-252 "Speech Synthesis" LB-54 "Circuit for Evaluation of Custom Vocabulary EPROM Prototype set" ( circuit for 2716 EPROMs ) AN-287 "Voice Recording Techniques for Speech Synthesis" ( not much about Digitalker )

One could put the question also in these two newsgroups: comp.robotics.misc comp.speech.research

I did think about that too. But: the chip is NMOS and utterly obsolete. The bitrate/quality

1-3kbit/sec was chosen low due to limited memory in 1980 ( 8kx8 ROMs ). People expect a less "robotic" sound today. To me it seems more appropriate to implement Mozer on an 8 bit controller (68HC908) with 60k flash in software. It should be easier too: reverse-engineering the MM45104 will be timeconsuming.

MfG JRD

Reply to
Rafael Deliano

Hi and thanks for your detailed and much appreciated reply; I will investigate doing Mozer on a more modern arch but since I have the old chips my experiments will continue. I will post any results when available.

Regards,

Michael

Reply to
msg

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.