Does speech synthesis support custom pronunciation or number reading?

Yes, speech synthesis supports custom pronunciation and number reading.

Custom Pronunciation

Speech synthesis systems often allow users to define how specific words or phrases are pronounced. This is useful for proper nouns (e.g., names, brands) or technical terms that may not follow standard phonetic rules. For example:

• A company name like "Xiaomi" can be customized to match its official pronunciation.
• A medical term like "COVID-19" can be adjusted for clarity.

In Tencent Cloud's speech synthesis service, users can upload custom pronunciation dictionaries to ensure accurate rendering of specific terms.

Number Reading

Speech synthesis can also be configured to read numbers in different formats. For example:

• "1234" can be read as "one thousand two hundred thirty-four" (default) or "twelve thirty-four" (abbreviated).
• Dates like "2024-05-20" can be read as "May twentieth, two thousand twenty-four" or "twenty twenty-four, May twentieth."

Tencent Cloud's speech synthesis supports flexible number formatting options, allowing users to choose between natural language or digit-by-digit reading based on their needs.

Example use case: A financial app may require numbers to be read in a specific format for clarity, while a news app might prefer natural language for better user experience.