Pronunciation rules and processing methods for text symbols in speech recognition and text processing.-Smart Science and Education Content Platform(SSECP)-阿里云帮助中心

Symbols supported within words

Symbol	Example	Description
[a-z][A-Z]	good	Common English words composed of the 26 letters of the alphabet (uppercase and lowercase) are supported.
-	four-legged	Hyphen within a word
'	don't	Hyphen within a word
.	John.Smith	Hyphens within words
:	18:16
@	--	Used in email addresses
_	--	Intra-word hyphen

Supported word boundary separators

Symbol	Example	Description
,	No, thank you.	Separator within a sentence
.	Good morning. Nice to see you.	Sentence separator
;	I like bread; I also like rice.	Separator within a sentence
?	How old are you?	Sentence separator
!	Have a nice day!	Sentence separator
"	"May I speak to Tom," I said.	Quotation mark
@	My email is mi@.**.	Email address separator. Pronounced as "at". Characters after the at sign are read as individual letters.
.	My email is mi@.**.	Email address separator. Pronounced as "dot".

Note: If a word ends with a period, the system first checks an abbreviation list. If the word is not an abbreviation, the period is treated as the end of a sentence.

Supported number pronunciations

Example	Pronunciation	Notes
2	two / february
56	fifty six
456	four hundred fifty six/four five six
1990	nineteen ninety/one thousand nine hundred ninety/one nine nine o
87th	eighty seven th
19:40	nineteen fourty/twenty to twenty
19.21	nineteen twenty one	Text that contains a period is split into two numbers. To pronounce the period as "point", manually replace "." with the word "point" in the text.
1,000	one o o o/one zero zero zero	A number that contains a comma is split into two words. To treat it as a single word, explicitly remove the comma.

Note: Numbers are generally not included in the internal dictionary. The scoring engine predicts pronunciation based on the rules above. For best results, convert numbers to their corresponding English words.

Handling of other common symbols

Symbol	Description	Example
Non-UTF-8 encoding	Converted to UTF-8 encoding.	This typically applies to the first 128 characters of the ASCII set.
Other symbols related to pronunciation, pauses, or speed	Replaced with a supported word boundary separator, such as a comma (,), semicolon (;), question mark (?), exclamation point (!), or quotation mark (").
Other symbols not related to pronunciation, pauses, or speed	Full-width letters, numbers, and punctuation are converted to their half-width equivalents. All other symbols are removed because they do not affect spoken language assessment.