Inworld acquires Ultravox, unifies AI voice agent platform and speech model
Inworld, an AI research lab and inference provider, has acquired Ultravox, a platform for building real-time voice AI agents. The acquisition combines Ultravox's conversational intelligence—understanding, reasoning, tool calling, and turn-taking—with Inworld's voice and realtime inference engine. Built-in Inworld voices on Ultravox now run on Realtime TTS-2, offering natural-language steering for tone, pace, and emotion, support for over 200 languages, and sub-100ms latency. The long-term goal is a single speech-to-speech system.
IllustrationEditorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page reads the event directly, while its address stays stable when the title changes.
- Summary covers the current reports
Cross-source coverage
Reporting timeline
Inworld acquires Ultravox, combining AI voice and agent platform in one company
Inworld, an AI research lab and inference provider, has acquired Ultravox, a platform for building real-time voice AI agents. The acquisition means one company now owns both the voice an AI agent speaks with and the platform the agent runs on. Ultravox handles understanding, reasoning, tool calling, and timing for voice agents used in support lines, tutoring, and companion apps. Inworld trains its own speech models and runs LLMs for other companies. As an immediate benefit, Ultravox customers get access to Inworld's TTS-2 text-to-speech model, which is top-ranked on the independent Artificial Analysis leaderboard. TTS-2 supports over 200 languages, starts speaking in under 100 milliseconds, and accepts natural-language direction for tone and emotion. The longer-term goal is speech-to-speech: a single system that listens, reasons, and speaks, rather than three separate steps.
Read sourceInworld acquires Ultravox to advance speech-to-speech AI voice agents
Inworld has acquired Ultravox, a platform for building voice agents capable of listening, reasoning, calling tools, and responding in live conversations with turn-taking and interruption handling. The combined entity aims to integrate speech understanding, reasoning, and expressive voice generation more closely for speech-to-speech agents. As part of the integration, built-in Inworld voices on Ultravox now run on Realtime TTS-2, with existing voice IDs remaining functional without migration or code changes. TTS-2 also adds natural-language steering capabilities to Ultravox agents, enabling more dynamic and context-aware voice interactions.
Read sourceInworld acquires Ultravox to build realtime speech-to-speech voice agent platform
Inworld has acquired Ultravox, a platform used by developers to build realtime voice agents. The integration combines Ultravox's conversational intelligence—understanding user intent, reasoning, tool calling, and managing turn-taking and interruptions—with Inworld's voice and realtime inference engine. The first upgrade is already live: built-in Inworld voices on Ultravox now run on Realtime TTS-2, offering natural-language steering for tone, pace, and emotion, support for over 200 languages, and sub-100ms first-audio latency. For existing users of Inworld voices on Ultravox, the upgrade is automatic with no migration, code changes, or additional cost. Voices from other providers remain unaffected. The end goal is a single realtime speech-to-speech system that understands, reasons, and speaks naturally.
Read sourceShow 2 older updatesHide older updates
Inworld acquires voice-agent platform Ultravox, moves voices onto Realtime TTS-2
Inworld, an AI company, has acquired Ultravox, a full voice agent platform. Ultravox understands speech and intent, calls tools, completes tasks, and handles turn-taking and interruptions in real time. Inworld has already moved Ultravox's built-in voices onto its Realtime TTS-2 model. Realtime TTS-2 accepts plain-language instructions, allowing agents to dynamically adjust pace, tone, and emotional delivery for specific speech segments, rather than setting a fixed rate for an entire call. For example, the agent can slow down while explaining a complex step and then return to normal speed.
Read sourceInworld acquires Ultravox to build realtime speech-to-speech AI voice agents
Inworld has acquired Ultravox, a platform used by developers to build realtime voice agents. The integration combines Ultravox's understanding and reasoning capabilities with Inworld's voice and realtime inference engine, creating a complete speech-to-speech loop: listen, understand, reason, act, respond. The first upgrade is already live, with built-in Inworld voices on Ultravox now running on Realtime TTS-2, offering natural-language steering for tone, pace, and emotion, support for over 200 languages, and sub-100ms first-audio latency. Users already using an Inworld voice on Ultravox receive the upgrade automatically with no migration, code changes, or additional cost. Voices from other providers remain unaffected. The endgame is a single realtime speech-to-speech system that understands, reasons, and speaks naturally.