some roman numbers are not well pronounced as roman numbers

jacobchawala commented 3 years ago

Is your feature request related to a problem? Please describe.

Some roman numbers are not well pronounced as roman numbers. they are pronounced badly as letters or words. such roman numbers that are badly pronounced include: I V VI X XXX XL L LI LII LIII LIV LV LVI LVIII LIX LX LXI LXII LXIII LXIV LXV LXVI LXVII LXVIII LXIX LXX LXXI LXXII LXXIII LXXIV LXXV LXXVI LXXVII LXXVII LXXVIII LXXIX LXXX LXXXI LXXXII LXXXIII LXXXIV LXXXV LXXXVI LXXXVII LXXXVII LXXXVIII LXXXIX XC XCI XCII XCIII XCIV XCV XCVI XCVII XCVIII XCIX C CC CCC CD D DC DCC DCCC DM M MM MMM

Describe the solution you'd like

they should be well pronounced starting with the word "roman" for example, roman fifty, roman sixty, roman seventy roman eighty, roman ninety, roman one hundred, roman two hundred, roman three hundred, roman four hundred, roman five hundred, roman six hundred, roman seven hundred, roman nine hundred, roman one thousand, roman two thousand, roman three thousand, ..... and so on.

Describe alternatives you've considered

Alternatively, NVDA speech dictionary should be improved to include those roman numbers that are badly pronounced.

Additional context

NVDA pronounce well some roman numbers. Such roman numbers that are well pronounced by NVDA as roman include: II III IV VII VIII IX XI XII XIII XIV XV XVI XVII XVIII XIX XX XXI XXII XXIII XXIV XXV XXVI XXVII XXVIII XXIX XXXI XXXII XXXIII XXXIV XXXV XXXVI XXXVII XXXVIII XXXIX XLI XLII XLIII XLIV XLV XLVI XLVII XLVIII XLIX

josephsl commented 3 years ago

Hi, is this with all speech synthesizers or with specific ones? If the latter, I advise creating custom voice dictionary entries. Thanks.

CyrilleB79 commented 3 years ago

Hi

This seems more something that is managed by speech synthesizer since all speech synthesizers do not pronounce these numbers the same way.

Anyway pronouncing these expressions as roman numbers is a trade-off between:

if these roman numbers are frequently used or not
if these expressions are also used for something else than roman numbers:
- For number consisting in only one letter, it is clear that a single letter may be used for something else than roman numbers. E.g. :
"Press CTRL + SHIFT + L to add a bullet list"
- Some two or three letter abbreviations may be commonly used. E.g:
- VI = Visually Impaired
- XL = extra large (T-shirt size)
- CD = Compact Disc
- DC = direct current
- and probably many other in more specific fields (CCC, DC, MM, etc.)

And it seems to me that synthesizers already try to find the best solution to solve this trade-off as much as possible.

It would be very helpful to know why you need all these combinations to be pronounced as roman number. Real-life examples would really help. I cannot imagine real life situations where roman number above 30 are used.

feerrenrut commented 3 years ago

Without extra semantics to identify these as roman numerals, it would be a mistake to assume they are. Visual users must identify these as roman numerals based on context.

As requested by @CyrilleB79 please do provide concrete examples of where you run into this as a problem, perhaps there are other solutions we can offer.

jacobchawala commented 3 years ago

Roman numbers to be pronounced correctly is very useful for students when learning how to write and pronounce roman numbers. I heard one student saying: "one role of NVDA is that it helps us pronounce words and numbers correctly. I hear writee a constitution of which there are articles with roman numbers I-L. that iw where I discovered this issue. I am using Espeak NG and I happened to use Eloquence synthesizers, but the issue is the same, not pronounced correctly some roman numbers.

On Wed, Feb 10, 2021 at 6:01 AM Reef Turner notifications@github.com wrote:

Without extra semantics to identify these as roman numerals, it would be a mistake to assume they are. Visual users must identify these as roman numerals based on context.

As requested by @CyrilleB79 https://github.com/CyrilleB79 please do provide concrete examples of where you run into this as a problem, perhaps there are other solutions we can offer.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/nvaccess/nvda/issues/12052#issuecomment-776404309, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEKBI32OPZ7TQ43FIMEEVXTS6HZG7ANCNFSM4XKDZU2Q .

-- +++ Yours Sincerely, Jacob Florian Chawala JFC

Phone No. 0787016381

Emails: jacobchawala@gmail.com jacobchawala@yahoo.com

Webmail: www.caict.netai.net

Skype ID: Jacob.Florian.Chawala +++

Adriani90 commented 3 years ago

This could be something that might be handled in a new punctuation category, as I proposed in #11467, a punctuation level for mathematical content or scientific content could be added where all the scientific symbols, greek letters and roman numbers could be pronounced by default according to symbols dictionary and speech dictionary. I guess this would be a category which would be used most often by students, and they would take the benefit of it. At least it would give some more confort in this regard.

If a new category is envisaged, then we could think of adding all the roman numbers to the speech dictionary and assign the mathematical cathegory to them.

feerrenrut commented 3 years ago

The problem is, pronouncing these characters as roman numerals, means that in other situations they will be pronounced incorrectly. For instance reading the news about the Chinese politician Xi Jinping would be pronounced as "eleven Jinping".

Perhaps in the future we could have an earcon that recognizes words / characters with multiple possible substitutions. The user could then explore a menu of the different options and decide which is most appropriate based on the context.

christopherJ-Defra commented 2 years ago

I think this is related, but more fixable, but may be down to the speech synthesiser not NVDA.

NVDA appears to read out UNICODE Roman Numerals incorrectly. For example Ⅲ (Roman Numeral Three) is read out as letter 2 1 7 2 This is extra odd as 2172 is lowercase ⅲ and the uppercase is 2162

I would expect Unicode characters to be read out correctly, and not require contextual interpretation (e.g. Xi Jinping) ⅩⅠ or Ⅹⅰ should not be confused (by software at least) as the same as Xi

Specifically relevant as "Queen Elizabeth Ⅱ" and "King Charles Ⅲ"

This is using "eSpeak NG" on Windows 10 Using "Microsoft Speech API version 5" or "Windows OneCore Voices" the character is not spoken at all

mgifford commented 1 year ago

This still seems to be a problem: https://shkspr.mobi/blog/2023/03/unicode-roman-numerals-and-screen-readers/

nvaccess / nvda