Resemble AI Dramabox: first open-source voice model with built-in cryptographic provenance signature. Expressive performance plus verifiable ownership. Available now.
Wispr AI is in talks to raise $260M at a $2B valuation, more than doubling, as its Wispr Flow voice dictation product attracts investor demand.
Mira Murati's Thinking Machines launches TML-276B: a native real-time interaction model that eliminates voice activity detection and redefines real-time AI standards.
OpenAI launched three voice models at once in its Realtime API: GPT-Realtime-2 brings GPT-5 reasoning; Translate and Whisper add streaming capabilities.
OpenAI releases GPT-Realtime-2 plus translation and transcription models in the Realtime API — a coordinated voice-modality push.
AssemblyAI's Universal-3 Pro Streaming uses an LLM as the decoder, achieving <300ms P50 latency, mid-sentence multilingual switching, and runtime vocabulary injection.
OpenAI's open-source realtime-voice-component React package lets developers add GPT-powered voice control to apps via narrow action definitions.
xAI's Grok TTS API achieves 5% word error rate vs ElevenLabs' 12%, now available via API at below-ElevenLabs pricing — a direct challenge to the leading voice AI provider.
Gemini 3.1 TTS launches with inline audio tag prompting — [whispers], [screams], [slow] — enabling expressive, controllable speech synthesis.
Curated AI insights — sent when there's something worth your inbox.