SSML (Speech Synthesis Markup Language)

SSML (Speech Synthesis Markup Language) is a W3C-standard XML-based markup language used to control how text-to-speech systems pronounce and render text. With SSML, you can insert pauses, control speaking rate, adjust pitch and volume, emphasise words, specify pronunciation for unusual terms (using IPA or phoneme spelling), switch voices mid-sentence, and add prosodic cues like rising or falling intonation. SSML is used by TTS providers including Amazon Polly, Google Cloud TTS, Microsoft Azure Speech, and many purpose-built platforms. A simple SSML example: `<speak>Hello <break time="500ms"/> and welcome.</speak>` produces "Hello [half-second pause] and welcome." SSML is particularly useful for applications where precise pronunciation matters — brand names, technical jargon, foreign-language words, phone numbers, and currency amounts. Most major TTS platforms accept SSML input via their APIs, though the set of supported tags varies by provider.

How it works

Core SSML tags include `<break>` for pauses, `<emphasis>` for stressed words, `<prosody>` for rate/pitch/volume control, `<phoneme>` for custom pronunciation via IPA, `<sub>` for text substitution (useful for pronouncing acronyms), `<say-as>` for interpreting input (phone numbers, dates, ordinals), and `<voice>` for switching voices mid-speech. Advanced SSML supports `<lang>` tags for inline language switching (useful for mixed Hindi-English code-switched content) and `<lookup>` for custom pronunciation dictionaries. Not all SSML tags work on all platforms — Google Cloud TTS supports a slightly different set than Amazon Polly. Modern preset-based TTS platforms like VoisLabs abstract SSML complexity behind named tone presets (horror, YouTube, devotional, etc.) — creators pick a preset instead of hand-writing SSML.

Examples

Related terms

Learn more

Frequently Asked Questions