SSML states that each sentence can be wrapped in an <s> </s> tag. For Google Assistant, doing this enormously improves the output of the TTS. Should SMD support this in some way? Detecting sentences might be hard (splitting on period is probably too naive); so the question is what is the most sensible way to implement this?
SSML states that each sentence can be wrapped in an
<s> </s>tag. For Google Assistant, doing this enormously improves the output of the TTS. Should SMD support this in some way? Detecting sentences might be hard (splitting on period is probably too naive); so the question is what is the most sensible way to implement this?