|
| From: | David Hill |
| Subject: | Re: [gnuspeech-contact] what's the status? |
| Date: | Fri, 3 Feb 2006 16:51:11 -0800 |
| Hi Robert, I have put three sound files up on my university web site. You can access them at the following URLs: (the University system substitutes "pages" for "www" -- either will actually do) The first was the first synthesis done using the tube-model based synthesis system on the NeXT, before the system was fully complete (most noticeably, the fricatives are absent, though you may be unaware of this for most of them because of the formant transitional cues plus background noise providing some faking). However, you will notice that the "s" in Christmas is really absent. It was produced under a deadline for a "Christmas Teaser" as part of the Trillium advertising as we were finishing the commercial version of the software. However, it is interesting because being sung, the rhythm and intonation are tightly determined. Also, my colleague Leonard Manzara, who did his PhD in music, with a technical thesis on acoustic reflections and imaging in rooms and halls, arranged the synthesis in four voice parts, and processed the output with his acoustic simulation software to put the "choir" in a large hall (30 meters square, if I remember correctly). It is quite melodious. The words are a little strange. "God and man this day are one, Even more than fife and drum; So these instruments we play, Tu-re-lu-re-lu, pat-a-pat-a-pan, So these instruments we play For a joyful Christmas day!" The second file is a spoken version of the Monty Python "Lumberjack" song (Oh! I'm a lumberjack and I'm OK ..." If you don't have access to the words, I can send them. The third one is a spoken rendering of a poem written by a Dutchman G.N. Trenite under the pseudonym "Charivarius". The poem was written to document the vagaries of English spelling & pronunciation and I chose to synthesis it because it was a pretty stringent test of the dictionary I was building at the time,it also tested the rhythm and intonation, and was of more interest than some word list (though it gets a bit list-like at times). In fact, the rhythm and intonation for the second two files is determined entirely from the text, and from the rules and data we derived from research at the U of Calgary. In the absence of any grammatical analysis, the algorithms have to rely on word structure and punctuation, plus what we know of English rhythm and intonation. This makes it impossible, for example, to distinguish "lead" (the metal) from "lead" (the act of providing inspiration to a group of people). The punctuation (normal picky punctuation) is really quite important to good synthesis. Even people find reading without punctuation can be difficult and it can change the meaning to get it wrong (e.g. A: We'll meet at 5 then? B: No, earlier. This means B can't meet as late as 5. B might have said B: No earlier. This would mean he couldn't meet any earlier. Just a simple example of punctuation and its relationship to rhythm and intonation. Trenite didn't cover such things as "Featheringstonehaugh" (a surname in Britain pronounced "fanshaw"), but he covered a lot. The word of the poem appeared in Ian Witten's book "Principles of Computer Speech" (Academic Press 1982 ISBN 0-12-760760-9) written about the time we were writing up much of our original research in which Ian participated. Unfortunately, the book doesn't really present our research results (I suppose it was off to press some time before the newest research was complete) but it does give some idea of where we were coming from. It has been out of date for some time now and I am thinking of writing a new book myself to bring things up-to-date. Ian's book is still one of the better sources available. For convenience, and because the version of the poem I used may be somewhat at variance with the version Ian quoted (I don't know which is truer to the original, which appeared in "Punch" -- a British humorous weekly -- in the early part of the last century), I provide, herewith, the words of the synthesised version. The chaos A poem on English pronunciation Charivarius, (G.N. Trenite: 1870—1946). Dearest creature in creation, Studying English pronunciation, I will teach you in my verse Sounds like corpse, corps, horse and worse. It will keep you, Susy, busy, Make your head with heat grow dizzy; Tear in eye your dress you'll tear. So shall I! Oh hear my prayer: Pray console your loving poet, Make my coat look new, dear, sew it. Just compare heart, beard and heard, Dies and diet, Lord and word. Sword and sward, retain and Britain, (Mind the latter, how it's written). Made has not the sound of bade, Say—said, pay—paid, laid, but plaid. Now I surely will not plague you With such words as vague and ague, But be careful how you speak: Say break, steak, but bleak and streak, Previous, precious; fuschia, via; Pipe, shipe, recipe and choir; Cloven, oven; how and low; Script, receipt; shoe, poem, toe. Hear me say, devoid of trickery; Daughter, laughter and Terpsichore; Typhoid, measles, topsails, aisles; Exiles, similes, reviles; Wholly, holly; signal, signing; Thames, examining, combining; Scholar, vicar and cigar, Solar, mica, war, and far. Desire—desirable, admirable—admire; Lumber, plumber; bier but brier; Chatham, brougham; renown but known, Knowledge; done, but gone and tone, One, anemone; Balmoral, Kitchen, lichen; laundry, laurel; Gertrude, German; wind and mind; Scene, Melpomene, mankind; Tortoise, turquoise, chamois-leather. This phonetic labyrinth Gives moss, gross; brook, brooch; ninth, plinth. Billet does not end like ballet; Bouquet, wallet, mallet, chalet; Blood and flood are not like food, Nor is mould like should and would. Banquet is not nearly parquet, Which is said to rime with darky. Viscous, viscount; load and broad; Toward, to forward, to reward. And your pronunciation's O.K. When you say correctly; croquet; Rounded, wounded; grieve and sieve; Friend and fiend, alive and live, Liberty, library; heave and heaven; Rachel, ache, moustache; eleven. We say hallowed, but allowed; People, leopard; towed, but vowed. Mark the difference moreover Between mover, plover, Dover; Leeches, breeches; wise, precise; Chalice, but police and lice. Camel, constable, unstable, Principle, discipline, label; Petal, penal and canal; Wait, surmise, plait, promise; pal. Suit, suite, ruin; circuit, conduit, Rime with: "shirk it" and "beyond it"; But it is not hard to tell Why it's pall, mall, but PallMall. Muscle, muscular; goal and iron; Timber, climber; bullion and lion; Worm and storm; chaise, chaos, chair; Senator, spectator, mayor. Ivy, privy; famous; clamour, And enamour rime with "hammer". Pussy, hussy and possess, Desert, but dessert, address. Golf, wolf; countenants; lieutenants Hoist, in lieu of flags, left pennants. River, rival; tomb, bomb, comb; Doll and roll, and some and home. Stranger does not rime with anger, Neither does devour with clangour. Soul, but foul; and gaunt, but aunt; Font, front, won't; want, grand and grant; Shoes, goes, does. Now first say finger, And then; singer, ginger, linger. Real, zeal; mauve, gauze and gauge; Marriage, foliage, mirage, age. Query does not rime with very, Nor does fury sound like bury. Dost, lost, post; and doth, cloth, loth; Job, Job; blossom, bosom, oath. Though the difference seems little We say actual, but victual; Seat, sweat; chaste, caste; Leigh, eight, height; Put, nut; granite but unite. Reefer does not rime with deafer, Feoffer does, and zephyr, heifer. Dull, bull; Geoffrey, George; ate, late; Hint, pint; senate, but sedate. Scenic, Arabic, Pacific; Science, conscience, scientific. Tour, but our, and succour, four; Gas, alas and Arkansas! Sea, idea, guinea, area, Psalm, Maria, but malaria. Youth, south, southern; cleanse and clean; Doctrine, turpentine, marine. Compare alien with Italian. Dandelion with battalion, Sally with ally, Yea, Ye, Eye, I, ay, aye, whey, key, quay. Say aver, but ever, fever, Neither, leisure, skein, receiver. Never guess—it is not safe; We say calves, valves; half, but Ralf. Heron, granary, canary; Crevice and device and eyrie; Face, preface, but efface, Phlegm, phlegmatic; ass, glass, bass; Large, but target, gin, give, verging; Ought, out, joust and scour, but scourging; Ear, but earn; and wear and tear Do not rime with "here" but "ere". Seven is right, but so is even; Hyphen, roughen, nephew, Stephen; Monkey, donkey; clerk and jerk; Asp, grasp, wasp; and cork and work. Pronunciation—think of psyche— Is a paling, stout and spikey; Won't it make you lose your wits, writing groats and saying "groats"? It's a dark abyss or tunnel, Strewn with stones, like rowlock, gunwale, Islington and Isle of Wight, Housewife, verdict and indict. Don't you think so, reader, rather Saying lather, bather, father? Finally: which rimes with "enough", Though, through, plough, cough, hough or tough? Hiccough has the sound of "cup", My advice is ... give it up! On Feb 2, 2006, at 11:20 PM, Robert Brewer wrote:
------- David Hill, Prof. Emeritus |---------------------------------------- | CS Dept, U. Calgary | Imagination is more | Calgary, AB T2N 1N4 Canada | important than knowledge. | OR address@hidden | Kill your television! | ------- David Hill, Prof. Emeritus |---------------------------------------- | CS Dept, U. Calgary | Imagination is more | Calgary, AB T2N 1N4 Canada | important than knowledge. | OR address@hidden | Kill your television! | |
| [Prev in Thread] | Current Thread | [Next in Thread] |