Text to speech - force short hand pronunciation for abbreviations such as I.e. and seq

Question

SnowMonkey90 Author

Level 1

8 points

Text to speech - force short hand pronunciation for abbreviations such as I.e. and seq

the title says it all.

the TTS pronounces the short hand as long hand, so i.e. becomes "that is" and seq becomes "the following"

now its not a huge deal in every day life but in science it is a pain.

RNA-seq is a technology used in my field but TTS pronounces it as "RNA the following"

and when you have sentences like "by bulk RNA-seq with no amplification (bulk SuperScript) and ~44% of genes detected by bulk RNA-seq" then it makes you want to die!

there has to be a way to tell the prediction program to "ignore abbreviations" of "say abrivaitions as the short hand"

please help, driving me mad

MacBook Pro with Retina display, OS X El Capitan (10.11.3)

Posted on Apr 30, 2016 7:43 AM

Reply

Answer 1

Apr 30, 2016 9:08 AM in response to SnowMonkey90

Play the following. Control sequences described in Apple's Speech Synthesis Programming Guide.

by bulk RNA- [[char LTRL]][[ctx WORD]] seq [[ctx NORM]][[char NORM]] with no amplification (bulk superscript) and ~44% of genes detected by bulk RNA- [[char LTRL]][[ctx WORD]] seq [[ctx NORM]][[char NORM]].

As you can see, it will require impractical measures of Speech Synthesis markup to avoid pitfalls in the speech results. Apple's TTS is unlikely to be worth the trouble to speak scientific, or even normal passages of text, in an acceptible manner.

Reply

Answer 2

SnowMonkey90 Author

Level 1

8 points

Apr 30, 2016 12:43 PM in response to VikingOSX

Thanks for the link and the info. Greatly appreciated.

Is there a way such as a sudo code, to force change the recognition of "seq" to run as "[[char LTRL]][[ctx WORD]] seq [[ctx NORM]][[char NORM]]"

I am not a coder so I may be way off here but is there a fine somewhere with the table that tells the voice what to play at any given instances?

Also, as I use the voice Samantha, which is owned by Nuance communications and they describe an ability to custem make filters in the code for stuff like this in there dev guide for Nuance.

Again, thanks for the help, my dyslexia thanks u

Reply

Answer 3

Apr 30, 2016 4:12 PM in response to SnowMonkey90

There is no pseudo code to replace the painful TTS markdown. Those embedded constructs are what the TTS engine recognizes as it is scanning the text to speak. There is no [[ say seq correctly, damit ]] as shorthand for that longer, more gnarly syntax. You might even have to respell seq as seekwince.

The voices and quality of voice can have a bearing on the recitation of your content.

The Apple Text-to-Speech engine is not based on Nuance technology, as it dates back to the early 1990s — so embedding Nuance markdown, if it exists, would not buy you anything, even if you are using a Nuance voice. Any Developer SDK from Nuance will be for the Windows Dragon Naturally Speaking client, and may require a Nuance server in the mix too.

Reply

Answer 4

SnowMonkey90 Author

Level 1

8 points

Apr 30, 2016 7:32 PM in response to VikingOSX

cool. i found a file that looks promising, /System/Library/PrivateFrameworks/SpeechDictionary.framework/Versions/A/Resourc es/HomophonesEng

the reason it looks promising is that the conversion of seq to "the following" is this type of phonetics/syntax, a homophone.

the problem is the file is locked, and i can not take ownership...

i get this:

--------:~ -------$ sudo chown -R --------- /System/Library/PrivateFrameworks/SpeechDictionary.framework/Versions/A chown: /System/Library/PrivateFrameworks/SpeechDictionary.framework/Versions/A/Resources/HomophonesEng: Operation not permitted

and then this is the lock it has on it...:

----------:~ -------$ ls -lO@e /System/Library/PrivateFrameworks/SpeechDictionary.framework/Versions/A/Resources/HomophonesEng -rw-r--r-- 1 root wheel restricted,compressed 88338 Sep 18 2015 /System/Library/PrivateFrameworks/SpeechDictionary.framework/Versions/A/Resources/HomophonesEng

i don't know how to override this, as i figure if i can add

     <key>seq</key>

     <string>šeq</string>

the šeq is the phonetics for what it should actually say... sort of... well its better than "the following" 🙂

but again, Xcode wont edit the file, save the file, as i have no rights...

maybe you or someone can help test this? i guess it would help allot of people if we can change the words to correct phonetics rather than rely on predicted phonetic assignment.

thanks a tone 🙂

Reply