Skip to main content
KugelAudio generates speech directly from text. There is no voice direction layer — you shape the output by how you write the input. This section covers every mechanism available to control pronunciation, pacing, and emphasis.

​
Supported controls at a glance

ControlSyntaxPage
Pauses<break time="300ms"/>, <break strength="medium"/>, <break/>Breaks
Speedspeed request parameter (0.8–1.2, whole request)Speed
Spell out characters<spell>text</spell>, <spell group="2">…</spell>Spell tags
Custom pronunciationInline IPA — /ˈkuːɡl̩/ — or pronunciation dictionariesPronunciation & IPA
Pacing & intonationPlain punctuation — see belowthis page
<break> and <spell> are the only tags processed in request text. Everything else — including SSML — is stripped before synthesis; see Unsupported tags.

​
Punctuation as pacing

The model respects natural punctuation cues — no special tags needed:
TechniqueEffect
, commaBrief pause between clauses
. periodSentence-end pause, falling intonation
… ellipsisLonger trailing pause
— em dashAbrupt pause / interruption feel
? question markRising intonation
! exclamationEnergetic delivery
\n newlineParagraph-level pause (similar to period)
Punctuation is the recommended way to add natural rhythm; reach for <break> tags when you need an explicit silence of a specific length (e.g. before a verification code).

​
Writing tips

  • Strip markdown before TTS. Asterisks, hashes, and bullet characters are read literally by the model.
  • No emoji. They are read out or garbled.
  • Write numbers as digits when they should be normalized (“You have 3 messages”) and always set language — see Text processing.
  • Keep sentences short and end them with punctuation — this also helps the streaming chunker start generation earlier (why).
  • !, ALL-CAPS, and ?! are prosody cues — the model will deliver them energetically. Use deliberately.

​
LLM system prompt pattern

When an LLM generates text that feeds directly into TTS, add instructions so it uses the supported controls correctly: 
You are a voice assistant. Format your responses for text-to-speech output:

- For email addresses and codes, use <spell> tags:
    "Your code is <spell>ABC-123</spell>"
- For a deliberate pause, use a break tag:
    "Your total is <break time="400ms"/> forty-two euros."
- Do NOT use markdown formatting (**, *, #, -, bullet points) — it will be read aloud literally.
- Do NOT use emoji.
- Do NOT use SSML tags other than <spell> and <break> — they are ignored.
- Keep sentences short. End with punctuation.
- Write numbers as digits when they should be normalized: "You have 3 messages."
For full voice-agent prompt design (turn-taking, error recovery, tool-call acknowledgements), see Voice Agent Prompting.

​
Unsupported tags

KugelAudio processes <spell>, <break>, and <prosody rate>. All other tags are silently stripped — the inner text is kept but the tag itself has no effect.
Tag / AttributeStatusAlternative
<speak> wrapperStrippedOmit — plain text is assumed
<prosody rate="...">Supported — per-span speedSee Speed
<prosody pitch="...">Rejected (400)No pitch control available
<prosody volume="...">Rejected (400)No volume control available
<emphasis>StrippedRephrase text for natural emphasis
<say-as interpret-as="...">StrippedUse <spell> for characters, normalization for numbers
<sub alias="...">StrippedWrite the spoken form directly, or use a dictionary
<phoneme>StrippedWrite inline IPA between slashes (/ˈkuːɡl̩/)
<audio>, <p>, <s>, <w>, <lang>Stripped—
Unknown tags are not validated at request time. Passing unsupported tags will not return an error — the tags are removed and the remaining text is synthesized. Test your output when migrating from a full-SSML provider like Google Cloud TTS, Amazon Polly, or Microsoft Azure.

​
Next steps

Breaks

Explicit pauses with break tags

Speed

The global speed parameter

Spell tags

Character-by-character pronunciation

Pronunciation & IPA

Fix how specific words are spoken