Inworld TTS 2
Inworld's most powerful and expressive text-to-speech model. Builds on TTS 1.5 with rich expressive speech, real-time latency, natural language steering (e.g. [whisper], [say excitedly]), and stronger multilingual support across 15 production languages plus 90+ experimental languages.
Quick start
# Inspect the price — a plain request returns the 402 challenge:
curl -i https://api.glianalabs.com/v1/infer \
-H "content-type: application/json" \
-d '{
"model": "inworld/tts-2",
"text": "<text to speak>"
}'
# Pay + run in one step with the mppx CLI (create a wallet: npx mppx account create):
npx mppx https://api.glianalabs.com/v1/infer \
-J '{"model": "inworld/tts-2", "text": "<text to speak>"}'Examples
Parameters
The text to be synthesized into speech. Maximum input of 2,000 characters.
When enabled, text normalization expands numbers, dates, times, and abbreviations before converting to speech. Turning this off may reduce latency.
Bits per second of the audio. Only for compressed audio formats (mp3, opus). The default is 128,000.
The output format for the audio. Supported formats are mp3, opus, wav, and flac. Defaults to mp3.
The synthesis sample rate in hertz. Accepts: 8000, 16000, 22050, 24000, 32000, 44100, 48000. The default is 48,000.
Speaking rate/speed, in the range [0.5, 1.5]. The default is 1.0. We recommend using values above 0.8 to ensure high quality.
Determines the degree of randomness when sampling audio tokens. Defaults to 1.0. Accepts values between 0 (exclusive) and 2 (inclusive). Higher values = more expressive, lower values = more deterministic.
Controls timestamp metadata returned with the audio. "word" returns word-level timing, "character" returns character-level timing. Note: adds latency. Defaults to none.
The ID of the voice to use for synthesizing speech. Defaults to Dennis.