Most of the focus in generative AI has been on text-based interfaces used to generate text, images, and more. The next wave appears to be voice, and it’s rolling in fast. In the latest development, Google today announced that it would be adding Chirp 3 — its speech-to-text and HD text-to-speech models — to its Vertex AI development platform starting next week.

Last week, Google quietly announced that Chirp 3 would be rolling out eight new voices for 31 languages. Use cases for the platform include building voice assistants, creating audiobooks, and developing support agents and voice-overs for videos. The news was announced at an event at Google’s DeepMind offices in London.

Its efforts are coming at the same time that others are leaping forward with their voice AI work. Last week, Sesame — the startup behind the viral, very realistic sounding “Maya” and “Miles” AI apps — announced the launch of its model for developers to build their own customized apps and services on top of its tech. Notably, there will be usage restrictions around Chirp 3 to try to keep a handle on misuse. “We’re just working through some of these things with our safety team,” said Thomas Kurian, CEO of Google Cloud, at a news event today.

ElevenLabs is among the major startups that have raised hundreds of millions in funding to expand their work in AI voice services. The news will bring Chirp 3 into the same stable as newer versions of its flagship LLM, Gemini, that are being tested, as well as its image-generation model Imagen and its pricey Veo 2 video generation tool. It remains to be confirmed whether what Google is releasing with Chirp 3 will be as “realistic” as some of the other AI efforts to create “human” voices (Sesame’s work stands out in particular). But as Demis Hassabis, the CEO of DeepMind, emphasized, this remains a marathon, not a sprint.

Google has been building “Chirp” voice services for years, going back to using the name as a code name for its early efforts to compete against Amazon’s Alexa service.