Spell-out Tag: Letter-by-Letter Pronunciation in Speech Synthesis

30-08-2023 , 05-09-2023

In SSML, the "spell-out" parameter for the "say-as" element lets you control how words are pronounced. It makes the speech synthesis system read the text letter by letter instead of saying the whole word.

Spell-out

This is used to instruct the system to pronounce each letter of a word individually. This parameter is helpful when you need to say complex or rarely used words so that the listener understands each letter clearly. If you want the system to pronounce the word "ASAP" letter by letter:

<say-as interpret-as="spell-out">ASAP</say-as>

Some narrators process the "Spell-out" parameter in speech synthesis differently, articulating each letter more distinctly.

<say-as interpret-as="spell-out">Ashlee</say-as>

Similarly, the "Characters" parameter functions in the same way.

Сharacters

<say-as interpret-as="characters">ASAP</say-as>

Example:

 <say-as interpret-as="characters">Ashlee</say-as>

Example:

Usage options

In speech synthesis using neural networks, the tags "spell-out" or characters" are suitable for the following situations:

• When you want to emphasize each letter of a word for clarity.
• For spelling out acronyms or abbreviations.
• To ensure listeners can write down a specific term correctly.
• When pronouncing individual characters in a sequence or string.
• For reading out alphanumeric codes.
• When individual characters have specific meanings, like DNA sequences or certain codes.

The abbreviation reads better with this tag. Listen to the difference. Announcer Guy reads first without the spell-out tag, then with the spell-out tag, example:

DNA. <say-as interpret-as="characters">DNA</say-as>