[DECtalk] Adding words to Dectalk dictionary
ebruckert Bruckert
edbruckert at gmail.com
Tue Jan 25 16:23:15 EST 2011
Starting at what version I can't remeber but DECtalk had a similar thing.
It had a part of speech parser that tried to guess which was right. It would
do "he refused the refuse."
We discovered that the original version had coding mistake and it didn't
work as well as later versions. The ultimate was to use the [:pron alt]
command to force it to swap the pronunciation. Without a huge dictionary to
obtain part of speech info it will never get to high levels of accurracy.
On Tue, Jan 25, 2011 at 3:55 PM, Raymond Grote <musicalman1 at comcast.net>wrote:
> Whoa, that list might help a lot. Eloquence even got that portion of the
> manual right when it was talking about the two different pronunciations of
> permit. I don't know how though, they were spelled exactly the same except
> one had a semicolon after the parenthesis and one had a period. Maybe
> Eloquence has a superhuman context clue interpreter (SHCCI)? Hahaha.
> I didn't really look at the other homographs but that list is very
> comprehensive.
>
> ----- Original Message -----
> From: "Jayson Smith" <ratguy at insightbb.com>
> To: "DECtalk Discussions" <dectalk at bluegrasspals.com>
> Sent: Monday, January 24, 2011 11:27 PM
> Subject: Re: [DECtalk] Adding words to Dectalk dictionary
>
>
> > Hi,
> >
> > I've found my old copy of the DECtalk PC reference manual, and pasted in
> > the
> > appendix on homographs, which is what we're after here. I think in some
> > cases the software is smarter than what is indicated in this text, but
> not
> > always. The DECtalk 4.0 denotation for the alternate pronunciation of a
> > homograph is a / in front of the word, which obviously would appear in
> > printed text. At any rate, here goes, hope it helps.
> > Jayson
> >
> > APPENDIX C
> > Homographs
> > Homographs are pairs of words which are spelled exactly the same
> > but are pronounced differently. These are often different in
> > terms of which syllable is accented. For example, if permit is a
> > noun, the accent is on the first syllable (permit); if, however,
> > the word is used as a verb, then the accent is on the second
> > syllable (permit). This often makes a great deal of difference
> > in understanding DECtalk when it is speaking such words in
> > connected discourse.
> > In earlier versions of DECtalk, the default form was always the
> > noun. In later versions of DECtalk, the default form is the more
> > frequent form of the two. In the event the alternate
> > pronunciation is needed, you may insert the correct phonetics
> > from the list below. You may also obtain the alternate
> > pronunciation bypreceding the word with a slash ("/"). For
> > example, the word sow 'to mend or make clothing' will be the
> > default pronunciation. The pronunciation of the same word sow
> > "female pig' is done by placing a slant immediately before the
> > word, e.g, /sow. DECtalk PC handles more homographs than any
> > previous version of DECtalk.
> > The Table below is a new and expanded list of the common
> > homographs of English with alternative pronunciations in
> > phonetic transcription. In those cases where DECtalk PC does not
> > chose the correct pronunciation, you can simply use slants or
> > else insert the correct one phonetically from the list below.
> > SPELLING PRIMARY ALTERNATE
> > abstract 'aebstraekt aebstr'aekt
> > abuse axby'uz axby'us
> > addict axd'ihkt 'aedihkt
> > advocate 'aedvaxkeyt 'aedvaxkaxt
> > affix 'aefihks axf'ihks
> > ally 'aelay axl'ay
> > alternate 'aoltrrnaxt 'aoltrrneyt
> > animate 'aenihmeyt 'aenihmaxt
> > annex 'aenehks axn'ehks
> > appropriate axpr'owpriyaxt axpr'owpriyeyt
> > arithmetic axr'ihthmaxtixk aerixthm'ehtixk
> > articulate aart'ihkyeleyt aart'ihkyelaxt
> > associate axs'owshiyeyt axs'owshiyaxt
> > attribute axtr'ihbyuwt 'aetrixbyuwt
> > august 'aogaxst aog'ahst
> > bass b'eys b'aes
> > 111
> >
> > baton baxt'aon b'aetaxn
> > close kl'owz kl'ows
> > combat kaxmb'aet k'aambaet
> > combine kaxmb'ayn k'aambayn
> > compact kaxmp'aekt k'aampaekt
> > complex k'aamplehks kaxmpl'ehks
> > compound k'aampawnd kaxmp'awnd
> > compress kaxmpr'ehs k'aamprehs
> > concert k'aansrrt kaxns'rrt
> > conduct kaxnd'ahkt k'aandahkt
> > confederate kaxnf'ehdrrixt kaxnf'ehdrreyt
> > confine kaxnf'ayn k'aanfayn
> > conflict k'aanflihkt kaxnfl'ihkt
> > conglomerate kaxnxgl'aamrixt kaxnxgl'aamrreyt
> > console k'aansowl kaxns'owl
> > construct kaxnstr'ahkt k'aanstraxkt
> > content k'aantehnt kaxnt'ehnt
> > contest k'aantehst kaxnt'ehst
> > contract k'aantraekt kaxntr'aekt
> > contrast k'aantraest kaxntr'aest
> > converse k'aanvrrs kaxnv'rrs
> > convert kaxnv'rrt k'aanvrrt
> > convict kaxnv'ihkt k'aanvihkt
> > coordinate kow'aordeneyt kow'aordixnaxt
> > decrease diykr'iys d'iykriys
> > defect daxf'ehkt d'iyfehkt
> > delegate d'ehlixgaxt d'ehlixg`eyt
> > deliberate daxl'ihbrraxt daxl'ihbrreyt
> > desert d'ehzrrt dixz'rrt
> > desolate d'ehselixt d'ehseleyt
> > diffuse dixf'yuws dixf'yuwz
> > digest d'ayjhehst dayjh'ehst
> > discharge dixsch'arjh d'ihscharjh
> > discount d'ihskawnt dihsk'awnt
> > dove d'owv d'ahv
> > duplicate d'uwplixkeyt d'uwplixkaxt
> > elaborate axl'aebrraxt axl'aebrreyt
> > estimate 'ehstixmeyt 'ehstixmaxt
> > excerpt 'ehksrrpt ehks'rrpt
> > excuse ixksky'uz ehksky'us
> > expatriate ehksp'eytriyaxt ehksp'eytriieyt
> > exploit ixkspl'oyt 'ehksployt
> > export ehksp'ort 'ehksport
> > extract ehkstr'aekt 'ehkstraekt
> > ferment frrm'ehnt f'rrmehnt
> > frequent fr'iykwixnt friykw'eynt
> > geminate jh'ehmixnaxt jh'ehmixneyt
> > graduate gr'aejhuweyt gr'aejhuwaxt
> > impact 'ihmpaekt ixmp'aekt
> > implant ihmpl'aent 'ihmplaent
> > import 'ihmport ihmp'ort
> > imprint 'ihmprihnt ihmpr'ihnt
> > incense ixns'ehns 'ihnsehns
> > 112
> >
> > incline ixnkl'ayn 'ihnklayn
> > increase ihnkr'iys 'ihnkriys
> > insert ihns'rrt 'ihnsrrt
> > insult ihns'ahlt 'ihnsaxlt
> > interchange 'ihntrrcheynjh ihntrrch'eynjh
> > intimate 'ihntaxmaxt 'ihntaxmeyt
> > invalid ixnv'aelixd 'ihnvaxlixd
> > just jhixst jh'ahst
> > lead l'iyd l'ehd
> > live l'ihv l'ayv
> > minute m'ihnixt mayn'uwt
> > miscount m'ihskawnt mihsk'awnt
> > misprint m'IsprInt m|spr'Int
> > misuse mixs'yuz mixs'yus
> > moderate m'aadrraxt m'aadrreyt
> > object 'aabjheht axbjh'ehkt
> > overrun 'owvrrrahn owvrrr'ahn
> > perfect p'rrfixkt prrf'ehkt
> > permit prrm'iht p'rrmiht
> > pervert prrv'rrt p'rrvrrt
> > polish p'aalihsh p'owlixsh
> > postulate p'aascheleyt p'aaschelaxt
> > predicate pr'ehdixkeyt pr'ehdixkaxt
> > predominate prixd'aamixneyt prixd'aamixnaxt
> > present priyz'ehnt pr'ehzaxnt
> > proceed praxs'iyd pr'owsiyd
> > produce praxd'uws pr'aaduws
> > progress pr'aagrehs praxgr'ehs
> > project pr'aajhehkt praxjh'ehkt
> > protest pr'owtehst prowt'ehst
> > read r'iyd r'ehd
> > reading r'iydixnx r'ehdixnx
> > rebel r'ehbel rixb'ehl
> > recall rixk'aol r'iykaol
> > recap riyk'aep r'iykaep
> > recess r'iysehs riys'ehs
> > record r'ehkrrd rixk'ord
> > recount riyk'awnt r'iykawnt
> > refill r'iyfihl riyf'ihl
> > refresh riyfr'ehsh r'iyfrehsh
> > refund riyf'ahnd r'iyfahnd
> > refuse rixf'yuz r'ehfyus
> > reject rixjh'ehkt r'iyjhehkt
> > relapse r'iylaeps rixl'aeps
> > relay r'iyley rixl'ey
> > remake r'iymeyk riym'eyk
> > rerun r'iy*rahn riy*r'ahn
> > research r'iysrrch riys'rrch
> > resume riy|z'uwm r'ehzaxmey
> > retake riyt'eyk r'iyteyk
> > rewrite riyr'ayt r'iy*rayt
> > segment s'ehgmixnt sehgm'ehnt
> > separate s'ehpaxreyt s'ehpaxraxt
> > 113
> >
> > sow s'ow s'aw
> > subject s'ahbjhehkt saxbjh'ehkt
> > sublet saxbl'eht saxbl'eht
> > subordinate saxb'ordenaxt saxb'ordeneyt
> > survey s'rrvey srrv'ey
> > suspect s'ahspehkt saxsp'ehkt
> > syndicate s'ihndixkixt s'ihndixkeyt
> > tear t'er t'ir
> > torment torm'ehnt t'ormehnt
> > transform traensf'orm tr'aensform
> > transplant traenspl'aent tr'aensplaent
> > transport traensp'ort tr'aensport
> > upset axps'eht 'ahpseht
> > use y'uwz y'uws
> > wind w'ihnd w'aynd
> > wound w'awnd w'uwnd
> >
> > ----- Original Message -----
> > From: <Dectalk at aol.com>
> > To: <dectalk at bluegrasspals.com>
> > Sent: Monday, January 24, 2011 10:33 PM
> > Subject: Re: [DECtalk] Adding words to Dectalk dictionary
> >
> >
> >> Hi guys! I'll be in charge of the Dectalk dictionary, but I don't have
> >> anything yet. If it will do phrases, this won't be a problem. But,
> I'm
> >> thinking that the dictionary is single word only. So I have no clue
> how
> >> to
> >> handle double pronunciations.
> >>
> >> One question I have is, is there a control character that appears as a
> >> space, that I could use as a command for an alternative pronunciation?
> >> I
> >> know
> >> in one Dectalk version it was control something, but I can't remember
> >> what
> >> it was.
> >>
> >> I would live to stay away from a visible command so it won't show up in
> >> a
> >> document. I have no clue how.
> >>
> >> I have another question. If E-voice still existed, I could let
> everyone
> >> know about the Dectalk SAPI 5 within hours. I miss E-voice to no end!
> >> So
> >> on the internet, what is the best way to get the word out about this
> >> SAPI
> >> 5?
> >>
> >>
> >> SNOOPI BOTTEN
> >>
> >>
> >> In a message dated 1/24/2011 9:37:25 P.M. Eastern Standard Time,
> >> musicalman1 at comcast.net writes:
> >>
> >> Well, I am not familiar with the Dectalk dictionary in any way, but
> >> those
> >> kinds of rules would not be easy to implement even if such a dictionary
> >> existed. I believe most synths that can pronounce words two different
> >> ways
> >> use context clues to figure out which pronunciation to use. For
> >> instance,
> >> read has two pronunciations, as you know. For example:
> >> you should read the book by Thursday.
> >> or
> >> The book should be read by Thursday.
> >> You could probably just put phrases in the dictionary, and changing
> >> their
> >> pronunciations accordingly. That is, if the Dectalk dictionary supports
> >> phrases, which it should although i'm just guessing here. This might
> work
> >> in
> >> the example you mentioned, as putting words in front of record would
> >> determine how it is pronounced. For example, you would never say, the
> >> re-cord. So it is safe to change this in the dictionary. you could
> >> probably
> >> do that for a lot of different word combinations and it would be pretty
> >> accurate. Adding all these rules would take a lot of time, and you
> would
> >> have to be careful with it. If the Dectalk dictionary supports context
> >> clues
> >> though, I wouldn't know how you would help it determine which
> >> pronunciation
> >> is correct.
> >> These are just my thoughts. I don't know how accurate this is, but I am
> >> pretty sure it is a good work-around if no other can be found.
> >>
> >> ----- Original Message -----
> >> From: "Blake Roberts" <beroberts at hughes.net>
> >> To: "'DECtalk Discussions'" <dectalk at bluegrasspals.com>
> >> Sent: Monday, January 24, 2011 8:31 PM
> >> Subject: [DECtalk] Adding words to Dectalk dictionary
> >>
> >>
> >>> Hi all,
> >>> Snoopi and I have been involved with the Dectalk SAPI project for
> >>> several
> >>> months now. I am happy to report that a SAPI Dectalk is nearing
> >>> reality.
> >>>
> >>> The project leader gave Snoopi the task of adding words to the Dectalk
> >>> dictionary. Specifically, enabling Dectalk to say the same word two
> >>> different ways, such as record. Does anyone have ideas on how the
> >>> Dectalk
> >>> dictionary can be modified so that Dectalk will know when to use
> >> different
> >>> pronunciations for the same word?
> >>>
> >>> I know Snoopi is on this list. I am writing on his behalf because
> >>> Snoopi
> >>> asked me to inquire of people familiar with Dectalk. Any advice would
> >>> be
> >>> appreciated.
> >>> Blake
> >>>
> >>>
> >>> _______________________________________________
> >>> DECtalk mailing list
> >>> DECtalk at bluegrasspals.com
> >>> http://jaybird.no-ip.info/mailman/listinfo/dectalk
> >>
> >> _______________________________________________
> >> DECtalk mailing list
> >> DECtalk at bluegrasspals.com
> >> http://jaybird.no-ip.info/mailman/listinfo/dectalk
> >>
> >>
> >
> >
> >
> --------------------------------------------------------------------------------
> >
> >
> >> _______________________________________________
> >> DECtalk mailing list
> >> DECtalk at bluegrasspals.com
> >> http://jaybird.no-ip.info/mailman/listinfo/dectalk
> >>
> >
> > _______________________________________________
> > DECtalk mailing list
> > DECtalk at bluegrasspals.com
> > http://jaybird.no-ip.info/mailman/listinfo/dectalk
>
> _______________________________________________
> DECtalk mailing list
> DECtalk at bluegrasspals.com
> http://jaybird.no-ip.info/mailman/listinfo/dectalk
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://bluegrasspals.com/pipermail/dectalk/attachments/20110125/ea085048/attachment.html>
More information about the Dectalk
mailing list