@rohanpaul_ai: Inworld has acquired voice-agent platform Ultravox and already moved its built-in voices onto Realtime TTS-2. Ultravox …
Summary
Inworld has acquired the real-time voice-agent platform Ultravox and migrated its built-in voices to Realtime TTS-2, which supports natural-language inline instructions for controlling pace, tone, and emotion during a conversation.
View Cached Full Text
Cached at: 10/01/26, 04:15 AM
Inworld has acquired voice-agent platform Ultravox and already moved its built-in voices onto Realtime TTS-2.
Ultravox is a full voice agent platform. It understands what you said and what you meant, works through it, calls tools, finishes the task, and handles the turn taking and interruptions in real time.
Realtime TTS-2, which now sits behind the built-in Inworld voices on Ultravox, takes direction in plain language. i.e. with inline instructions you can change pace, tone and emotional delivery for a specific stretch of speech.
So instead of setting a rate to 0.85 for the whole call, the agent can ask the voice to slow down while it walks someone through a complicated step, then go back to normal.
Inworld AI (@inworld): We’re excited to announce that @ultravox_dot_ai is now part of Inworld.
Ultravox is the platform developers use to build real-time voice agents. We’ve worked with the team for a while through our TTS partnership, and today members of the team that built it are joining Inworld to
Similar Articles
@inworld_ai: Realtime TTS-2 is GA today, now the #1 model on Artificial Analysis and fastest TTS in the world of its class. The rese…
Inworld AI releases Realtime TTS-2 as generally available, claiming it to be the top model on Artificial Analysis and the fastest TTS in its class, representing a significant advancement in the company's history.
@rohanpaul_ai: I wasn't expecting Soniox TTS v2 (a text-to-speech model) to sound this natural. They just released this TTS v2 > Reall…
Soniox TTS v2 is a new text-to-speech model offering premium voice quality, expressive control via audio tags, high-fidelity voice cloning, support for 60+ languages, and low-latency streaming, priced at $0.70 per generated hour.
@rohanpaul_ai: A really cool real-time video models for interactive characters and worlds just launched. A1 doesn't generate a voice a…
Vivix.AI launched A1, a real-time video model for interactive characters and worlds that generates audio and video together in small consecutive segments, enabling seamless interruption and live, long-form conversation with persistent state.
@rohanpaul_ai: I think voice is about to become the default input layer for a lot of software. For me, AI is really making the keyboar…
The post discusses how AI is making voice the default input layer for software and introduces the Multilingual Voices feature to maintain business character across languages.
Advancing voice intelligence with new models in the API
OpenAI has announced three new voice models in its API: GPT-Realtime-2 with advanced reasoning, GPT-Realtime-Translate for live multilingual translation, and GPT-Realtime-Whisper for streaming transcription, aiming to enable more natural and action-oriented voice applications.