inworld/realtime-tts-1.5-mini

Ultra-fast, cost-efficient realtime text-to-speech with ~120ms latency and 15-language support

104.1K runs

Inworld Realtime TTS 1.5 Mini is Inworld’s ultra-fast, most cost-efficient realtime text-to-speech model. With ~120ms median latency and support for 15 languages, it’s built for real-time conversational AI at scale.

Ranked #1 on Artificial Analysis, Inworld Realtime TTS delivers natural, expressive speech at a fraction of the cost of alternatives.

Key features

  • ~120ms median latency: Built for real-time conversation and voice agents
  • 15 languages: English, Chinese, Japanese, Korean, Russian, Italian, Spanish, Portuguese, French, German, Polish, Dutch, Hindi, Hebrew, and Arabic
  • Emotion control: Add emotion markups like [happy], [sad], [angry] to control delivery
  • Non-verbal sounds: Insert [laugh], [sigh], [cough] and other vocalizations
  • SSML pauses: Use <break time="1s" /> to insert natural pauses
  • Multiple formats: MP3, WAV, OGG Opus, and FLAC output

Voices

Voice Description
Ashley A warm, natural female voice
Dennis Middle-aged man with a smooth, calm and friendly voice
Alex Energetic and expressive mid-range male voice, with a mildly nasal quality
Darlene Soothing, comforting Southern female voice, ideal for bedtime stories and narrations

Set voice_id to one of these preset voices. Custom voices you clone in your own Inworld account are scoped to that account’s API key and aren’t usable here.

Audio markups

The model supports rich text markups for expressive speech:

  • Emotions: [happy], [sad], [angry], [surprised], [fearful], [disgusted]
  • Delivery styles: [laughing], [whispering]
  • Non-verbal sounds: [breathe], [clear_throat], [cough], [laugh], [sigh], [yawn]
  • Pauses: <break time="1s" />, <break time="500ms" />

Choosing between Inworld Realtime TTS models

  • Realtime TTS 1.5 Mini: Ultra-fast (~120ms), most cost-efficient — best for high-volume, latency-sensitive applications
  • Realtime TTS 1.5 Max: Best balance of quality and speed (<200ms) — best for applications where voice quality is the top priority
Model created
Model updated