Speech Modes for Chess, Chat etc.
Posted: Tue Sep 08, 2009 4:57 am
I have been using Cepstral for only a week and find it quite good. Here are some suggested improvements.
It would be helpful if there were different speech modes for Cepstral voices that could be activated through user selection or from text style recognition in different applications.
I am using Cepstral William with BabasChess with the text-to-voice plug in so that I can listen to FICS (free internet chess server) lectures while I watch the game play. This has been great because I am not missing moves when there is a lot of text and vice versa.
Unfortunately Cepstral have a standard text substitution - if the letter b is followed by a number it interprets the b to mean "billion". So, when a night is moved to square b5 this is read as "billion five".
Chess has a standard nomenclature. It would be helpful if there was a chess mode for Cepstral voices to interpret this nomenclature:
Nomenclature.....Cepstral Reads.....Expected Speech
Nb5...................N billion five...........Night b5
QxRa1...............Q X R A 1...............Queen takes rook on a1
dxc....................D X C....................d takes c
Qa4...................Q A 4.....................Queen A4
Ra4+.................R A 4 plus...............Rook A4 check
Ba4++...............B A 4 plus plus........Bishop A4 checkmate
O-O....................O O.......................Castle King Side
O-O-O.................O O O....................Castle Queen Side
Kc5?...................K C 5......................King C5, Huh?
a3??....................A 3.........................A3, Ouch!
a4xb5!.................A 4 X B 5................A4 takes B5, Great
Bxg7!!.................B X G 7...................Bishop takes G7, Outstanding!
The "billion five" is an example that Cepstral is already making text substitutions. If it could make different substitutions for different situations that would be great. An alternate theme to chess may be chat-room, where the standard abbreviations like
LOL (Laugh out loud),
brb (be right back),
thx (thanks),
gg (good game),
yw (you're welcome) etc.
could be interpreted.
Other standard abbreviations are already substituted e.g. Rd., Ave., Mr, Mrs, Ms, Dr etc. However in some cases the abreviation could have different meanings or Cepstral does not recognise the abreviation. It would be great if the user could select the appropriate meaning or add their own abreviations.
e.g. ST could mean "saint" or "street", as in St. John or Apple St.. Maybe you could tell the difference if the "ST" comes before or after the noun. Similarly DR could mean "Doctor" or "Drive".
e.g. BSc (Bachelor of Science), APQP (Advanced Product Quality Planning), PPAP (Pre-Production Approval Process), AAAA (Association against Acronyms and Abbreviations )
Other substitutions:
When text is written as though pluralisation is dependant upon the circumstances e.g. "Read the word(s) aloud" then this should be read as though it were plural and the brackets should be ignored. Currently Cepstral reads this as "Read the word S aloud".
When reading acronyms it would be great to be able to configure Cepstral to always read capitals of unrecognised words as individual letters or to try and say the acronym phonetically. e.g. I work with bills of materials which is abbreviated to BOM. This is always pronounced BOMB and not read as B O M.
Lastly, it would be helpful if Cepstral could emphasise text in bold (strong voice), underline (Strong voice with pause either side of the underlined text), italic (inflection) or capitals (shout) as well as sentences ending in a question mark being formed as a question.
It would be helpful if there were different speech modes for Cepstral voices that could be activated through user selection or from text style recognition in different applications.
I am using Cepstral William with BabasChess with the text-to-voice plug in so that I can listen to FICS (free internet chess server) lectures while I watch the game play. This has been great because I am not missing moves when there is a lot of text and vice versa.
Unfortunately Cepstral have a standard text substitution - if the letter b is followed by a number it interprets the b to mean "billion". So, when a night is moved to square b5 this is read as "billion five".
Chess has a standard nomenclature. It would be helpful if there was a chess mode for Cepstral voices to interpret this nomenclature:
Nomenclature.....Cepstral Reads.....Expected Speech
Nb5...................N billion five...........Night b5
QxRa1...............Q X R A 1...............Queen takes rook on a1
dxc....................D X C....................d takes c
Qa4...................Q A 4.....................Queen A4
Ra4+.................R A 4 plus...............Rook A4 check
Ba4++...............B A 4 plus plus........Bishop A4 checkmate
O-O....................O O.......................Castle King Side
O-O-O.................O O O....................Castle Queen Side
Kc5?...................K C 5......................King C5, Huh?
a3??....................A 3.........................A3, Ouch!
a4xb5!.................A 4 X B 5................A4 takes B5, Great
Bxg7!!.................B X G 7...................Bishop takes G7, Outstanding!
The "billion five" is an example that Cepstral is already making text substitutions. If it could make different substitutions for different situations that would be great. An alternate theme to chess may be chat-room, where the standard abbreviations like
LOL (Laugh out loud),
brb (be right back),
thx (thanks),
gg (good game),
yw (you're welcome) etc.
could be interpreted.
Other standard abbreviations are already substituted e.g. Rd., Ave., Mr, Mrs, Ms, Dr etc. However in some cases the abreviation could have different meanings or Cepstral does not recognise the abreviation. It would be great if the user could select the appropriate meaning or add their own abreviations.
e.g. ST could mean "saint" or "street", as in St. John or Apple St.. Maybe you could tell the difference if the "ST" comes before or after the noun. Similarly DR could mean "Doctor" or "Drive".
e.g. BSc (Bachelor of Science), APQP (Advanced Product Quality Planning), PPAP (Pre-Production Approval Process), AAAA (Association against Acronyms and Abbreviations )
Other substitutions:
When text is written as though pluralisation is dependant upon the circumstances e.g. "Read the word(s) aloud" then this should be read as though it were plural and the brackets should be ignored. Currently Cepstral reads this as "Read the word S aloud".
When reading acronyms it would be great to be able to configure Cepstral to always read capitals of unrecognised words as individual letters or to try and say the acronym phonetically. e.g. I work with bills of materials which is abbreviated to BOM. This is always pronounced BOMB and not read as B O M.
Lastly, it would be helpful if Cepstral could emphasise text in bold (strong voice), underline (Strong voice with pause either side of the underlined text), italic (inflection) or capitals (shout) as well as sentences ending in a question mark being formed as a question.