Talk:TalknType

From OLPC
Revision as of 19:04, 11 January 2008 by 172.202.220.107 (talk)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

It would be awesome of the pronouncer for the word could "chunk" as is used to teach reading. See the links on the Phonics page for some resources. Maybe the text to speech gadget has a mode for this. This is a great activity idea. We need more activities for the youngest target age groups.

That looks really interesting... As a kid who was lucky enough to have a Speak&Spell whilst growing up, one of the most entertaining things about it was that it was obviously imperfect... There was a sense that, even as a kid who couldn't spell well, I could say the words better than the computer could, which meant that half the fun was in trying to decode what on earth the speech synthesiser was attempting to utter! There are many issues that arise here - the original Speak&Spell would request words like "for" and "four", which sound the same (are homophonic? is that the right word?)... To overcome this, the software had two sets of data - the description of the word that was asked for, and the spelling of that word. For example, the machine would say "Spell for, as in for you", or "Spell four, as in number four". I think it will be a while before the speech synthesiser is finessed enough to reliably split a word into perfect phonics, but that may not be such a bad thing... --Tomhannen 16:45, 3 January 2008 (EST)
Note: espeak definitely has support for phonetics—that is how it knows how to pronounce the words it is given. You can give it direct phonetic input, and the "-x" command line parameter tells it to output the phonemes of a word. I don't think there's an easy way to get the correspondences between the word parts and the phonemes, but perhaps some heuristics can be developed. —Joe 19:14, 4 January 2008 (EST)

Video

Tom, the video looks awesome! Where does one get the sugar-speechd daemon? —Joe 19:14, 4 January 2008 (EST)

Hi - thanks - the sugarspeechd can be built from [here], but I couldn't "make" it successfully on my olpc, so one of the screenreader project team sent me the binary directly which worked well. If you email me, I can send it to you, or try making it from the C code, and see how you go! I've requested hosting for the project on dev.laptop.org but haven't heard back from them yet... --Tomhannen 06:21, 5 January 2008 (EST)
The sugarspeechd is now part of the git page [here.]

word list

knight, lamb, mnemonic, phlegm, pneumonia, rhetoric, diarrhea, phoenix, damn, vaccuum, Iraq, weird, cholera, botulism, answer, honest, island, subtle, inn, giraffe, vineyard, asthma, psychology, feather, tsunami, Worcester, Chicago, Illinois, Michael, Phoebe, Berkeley, whine, whole, huge, Matthew, Paul, gradual, gnu, mane, main, Maine, parachute, cello, chord, enough, laugh, of, get, danger, scissors, iron, sugar, tissue, schism, schedule, schist, school, thyme, liquor, mosquito, linguistics, salmon, species, illusion, barbeque, pterodactyl, ctenoid, ghost, acquaint, biscuit, diaphragm, hymn, tongue, rhyme, sapphire, savvy, breathe, scene, flaccid, juice, horse, hoarse, conscience, ocean, schmooze, crescendo, fajita, onion, hallelujah, tortilla, choir, deceit, people, amoeba, hygiene, counterfeit, sovereign, carriage, mileage, luminous, who, flute, soup, true, lose, fruit, maneuver, canoe, through, two, courier, should, cocaine, arraign, straight, veil, beige, reign, eight, ballet, dossier, lingerie, reggae, porpoise, foreign, beryl, soul, foe, brooch, beau, oh, sew, mauve, pharaoh, furlough, cleanse, heifer, jeopardy, friend, lieutenant, bury, guess, plaid, meringue, bureaucracy, yacht, sausage, cough, Christ, height, aisle, isle, sign, geyser, dye, die, guide, sergeant, bazaar, stationery, stationary, heir, hair, hare, aerial, myrrh, grammar, colonel, queue, debt, fjord, siege, burqa, squirrelled, schlepped, strengths, aqueous, sequoia, aa, Io, Albuquerque, synchrocyclotron, broccoli, chaos