Standardized Audio Tags
Write expressive audio cues once and let SpeechSDK translate them for each provider.
Every TTS provider has its own way of handling expressive cues — ElevenLabs uses bracket syntax, Cartesia uses SSML, and most providers don't support them at all. SpeechSDK gives you a single, standardized format that works across all providers.
Write [tag] in your text and SpeechSDK handles the rest: passing tags through natively where supported, converting to SSML where needed, and cleanly stripping them (with warnings) everywhere else. One syntax, every provider.
import { generateSpeech } from "@speech-sdk/core"
const result = await generateSpeech({
model: "elevenlabs/eleven_v3",
text: "[laugh] Oh that is so funny! [sigh] But seriously though.",
voice: "voice-id",
})
console.log(result.warnings) // undefined — eleven_v3 supports all tagsProvider Behavior
| Provider | Behavior |
|---|---|
ElevenLabs (eleven_v3) | All [tag] passed through natively |
Cartesia (sonic-3) | Emotion tags ([happy], [sad], [angry], etc.) converted to SSML; [laughter] passed through; unknown tags stripped |
| All others | Tags stripped and warnings returned |
Warnings
When a provider doesn't support audio tags, SpeechSDK strips them from the text before sending the request and returns warnings in result.warnings:
const result = await generateSpeech({
model: "openai/gpt-4o-mini-tts",
text: "[laugh] Hello world",
voice: "alloy",
})
console.log(result.warnings)
// ["Audio tag [laugh] is not supported by openai/gpt-4o-mini-tts and was removed."]If the provider supports all tags, result.warnings is undefined.