Inworld TTS Paralinguistic Tags Don't Work — Here's What Does

Inworld's TTS-1.5 Max model ignores common paralinguistic tags like `[laugh]` and `[sigh]`, rendering them as silence or literal text instead of expressive audio. Developers at HoneyChat, a Telegram-native AI companion, tested multiple tag variants across 26 archetypes and 15 languages before discovering that SSML `<break>` tags, ellipsis-based pauses, and spelled-out onomatopoeia like "ha-ha" reliably produce the intended vocal effects.

If you've worked with expressive TTS in the last year you've probably seen the pattern: She paused. sigh "Fine, you can come in." Inline paralinguistic tags. Half the model demos use them. So when we wired up Inworld TTS-1.5 Max for HoneyChat https://honeychat.bot/ — Telegram-native AI companion where voice messages are a first-class output — we sprinkled laugh , sigh , breathe through the prompts and shipped. The audio sounded fine. Just… exactly the same as before. No laugh. No sigh. The tags were getting read out as silence at best, and as the literal text "sigh" at worst, depending on the voice. We tested all the variants we could find. None of them moved the needle. HoneyChat voice stack at a glance: en, ru, ja, zh, ko, es, fr, de, it, pt, pl, hi, ar, he, nl . voiceId strings in config/archetype voice ids.json . Generated via the Voice Design API and managed with core/voice design.py . core/voice clone manager.py — persistent voiceId minted from a WAV/MP3 sample. core/voice cache.py . VOICE GENDER MALE / VOICE GENDER FEMALE , not "male" / "female" strings. Passing the strings 400s silently.Tried on the same sentence, same voice, side-by-side audio comparison: | Pattern | What it did | |---|---| laugh sigh | Silence in output | laughs sighs | Sometimes read literally | laughs sighs | Silence asterisks get stripped | <laugh/ <sigh/ | Silence not valid SSML on Inworld | <emotion laugh</emotion | Silence | The Inworld API does not document support for any of these. We had assumed because every other TTS post on the internet uses them that they were a universal convention. They are not. What Inworld does expose is temperature and speakingRate After enough A/B-ing across 26 archetypes × 15 languages, four patterns reliably change the audio output. "You did what? " The asterisks get stripped from the spoken text but the emphasised word lands with audible stress. Works in every voice we tried. The cheapest, highest-hit-rate marker. "Fine... you can come in." Three dots produces a real pause with a tonal drop — the voice equivalent of a sigh, without trying to fake sigh . Five dots for a longer pause. The model interprets them as prosodic cues. <break for hard pauses <speak She paused. <break time="0.4s"/ "Fine, you can come in." </speak Inworld accepts a useful subset of SSML, and <break is the one that matters most for expressive speech. 0.2s for a beat, 0.4s for a sigh-pause, 0.8s for a beat-before-a-line-delivery moment. Wrap the whole text in <speak and the parser handles it. "Mmm... ha-ha, you're right." "ahh... I needed that." The model will render ha-ha , mmm , ahh , oh , nnn as the actual sound, because they're spellings of sounds rather than meta-tags. They sound far more natural than a synthesised laugh even when one exists. For emotional/intimate scenes, rhythmic repeats ah... ah... ah carry actual prosody. We use this for breath patterns where another TTS would want a breathe marker. In core/voice.py we run every chunk through enrich for tts line ~772 before handing it to Inworld. Regex-based, language-aware, idempotent: php def enrich for tts text: str, lang: str = "en" - tuple str, dict : """Return preprocessed text, request params . Strips fake paralinguistic tags, adds SSML breaks where appropriate, and bumps temperature/speakingRate for high-emotion scenes.""" text = STRIP FAKE TAGS.sub "", text text = ELLIPSIS TO BREAK.sub r'<break time="0.3s"/ ', text if "<break" in text: text = f"<speak {text}</speak " params = detect mood params text, lang return text, params The mood detector looks for emotional cues intensity words, repeated punctuation, onomatopoeia density and bumps temperature and speakingRate for the more expressive scenes. Same model, same voice, much more dynamic output, all without any inline tag that the model would have ignored. laugh / sigh is universal. sigh that emits silence looks identical to one that emits a sigh in any log. temperature , speakingRate , and a useful subset of SSML — not inline tags. "ahh..." is a thing the model can read; sigh is a meta-instruction it can't.The audio quality jump from these four patterns is meaningful — users notice. The cost is a 30-line preprocessor and the courage to delete every laugh your team has been sprinkling for months. This is from production work at HoneyChat — Telegram-native AI companion where voice messages are a first-class output. Canonical version: — HoneyChat Engineering temperature , speakingRate , SSML subset, voice design API. <break , <speak , prosody elements.