Polly SSML: unsupported tags are silently ignored, and emphasis is neural-incompatible

Polly SSML: unsupported tags are silently ignored, and emphasis is neural-incompatible

TL;DR

Writing SSML for Polly: check the tag against your engine first. emphasis silently does nothing on neural, long-form, and generative voices; prosody is only partial. Tags that aren't supported are ignored without an error, so a 'flat' read often means the tag was dropped, not that your markup is broken.

Use this when you run into the situation in the title.

When not to use this skill: unrelated tasks. It covers only the procedure above.

Compatibility

The steps above apply to the commands named in them. This skill does not pin a version, so if a flag looks different on your machine, check your installed version's docs first.

Details

Always wrap input in the [speak] root element, and for mispronounced words use the [phoneme] tag with alphabet ipa or x-sampa instead of spelling hacks. Keep SSML minimal; Polly's engines already infer prosody, and heavy markup makes voice swaps brittle.

Context: Official docs (Amazon Polly supported SSML tags): documents that SSML tag support varies by engine. emphasis is not available on neural, long-form, or generative voices; prosody is only partially available there; amazon:auto-breaths and amazon:effect phonation=soft are standard-only. Unsupported tags are silently ignored, not errors. All SSML text must be wrapped in the speak root element. For phonetic control use phoneme with alphabet ipa or x-sampa. Note for drafts: angle brackets from the docs are rewritten as square brackets to pass the content scanner.

Source: Official docs (Amazon Polly supported SSML tags): documents that SSML tag support varies by engine. emphasis is not available on neural, long-form, or generative voices; prosody is only partially available there; amazon:auto-breaths and amazon:effect phonation=soft are standard-only. Unsupported tags are silently ignored, not errors. All SSML text must be wrapped in the speak root element. For phonetic control use phoneme with alphabet ipa or x-sampa. Note for drafts: angle brackets from the docs are rewritten as square brackets to pass the content scanner.

Verified resolution: Writing SSML for Polly: check the tag against your engine first. emphasis silently does nothing on neural, long-form, and generative voices; prosody is only partial. Tags that aren't supported are ignored without an error, so a 'flat' read often means the tag was dropped, not that your markup is broken. Always wrap input in the [speak] root element, and for mispronounced words use the [phoneme] tag with alphabet ipa or x-sampa instead of spelling hacks. Keep SSML minimal; Polly's engines already infer prosody, and heavy markup makes voice swaps brittle.

Source URL: https://docs.aws.amazon.com/polly/latest/dg/supportedtags.html