Google Cloud Adds Chirp 3 Audio Generation to Vertex AI

Text-based interfaces for creating text, graphics, and other content have received the majority of attention in the field of generative AI. Voice seems to be the next wave, and it’s coming in quickly. Google recently announced on Monday that it is integrating the Chirp 3 artificial intelligence (AI) speech-to-text and HD text-to-speech models into its Vertex AI development platform. After a private preview, the model will now be available to all users of the AI platform. Chirp 3 is an audio generation model that adds a variety of custom voices with human-like intonations and inflections. 

Google secretly revealed last week that the upgraded model with eight new voices in 31 languages  would be added to Chirp 3. The platform may be used to create voice assistants, audiobooks, support agents, and video voiceovers, among other applications. The announcement was made during a gathering at Google’s London DeepMind headquarters.

Google Cloud said in a press release that the Chirp 3 AI model is now accessible through its Vertex AI platform. The news was revealed during the “Gemini for the United Kingdom” event, which took place at Google DeepMind’s London offices. The company’s Cloud business also made a number of other AI-related announcements at the event.

Beginning next week, Chirp 3 will be accessible on Vertex AI, joining other cutting-edge models like Veo, Imagen, and Gemini. With eight distinct speaker options and 248 distinct voices, Chirp 3’s HD Voices functionality will be accessible in 31 languages at launch. According to the business, Chirp 3 provides personalized speech production with emotional inflections and human-like intonation.

The business markets the AI model as a tool for use cases including sentiment analysis from customer conversations, audiobooks, podcast narration, real-time meeting transcription, and voice annotation. Additionally, it may be utilized to create speech-activated AI agents and voice assistants.

Its efforts coincide with those of others who are making rapid progress in speech AI. The business that created the popular, incredibly lifelike “Maya” and “Miles” AI apps, Sesame, revealed last week that its model for developers to create their own unique apps and services on top of its technology was now available.

According to TechCrunch, At a press conference today, it was pointed out that in an effort to prevent abuse, use constraints will be implemented around Chirp 3, the AI model will include some usage limitations in order to reduce the dangers of abuse. According to reports, Thomas Kurian, the CEO of Google Cloud, also stated during the event that the business is working on the safety aspects before making the model more widely available.

Google’s developer-focused cloud-based end-to-end platform for creating, implementing, and scaling AI models is called Vertex AI. The platform provides a single experience for developing new applications based on these models as well as for deploying and testing new models. The tech giant provides free credits to try the platform once, but a membership is needed to continue using it.

Several significant businesses, like ElevenLabs, have funded hundreds of millions of dollars to advance their work in AI voice services.

The announcement will place Chirp 3 in the same stable as its expensive Veo 2 video generating tool, Imagen, its image-generation model, and newer iterations of its flagship LLM, Gemini, which are undergoing testing.

Whether Google’s Chirp 3 will be as “realistic” as some of the other AI attempts to produce “human” voices—Sesame’s work stands out in particular—has not yet been verified. However, DeepMind CEO Demis Hassabis stressed that this is still a marathon rather than a sprint. 

The notion that artificial intelligence (AI) would solve all problems in the next few years is not something I see occurring anytime soon. Think that something akin to AGI won’t happen for a number of years,” he stated. In the medium to long term, it will alter things over the course of the following ten years. It’s one of those fascinating times in history.

In 2021, Google introduced Vertex AI, a platform that allows developers to create cloud-based machine learning applications. Naturally, there was a long time before the rise in interest in artificial intelligence (AI), and more especially generative AI, that coincided with the introduction of OpenAI’s GPT services.

In an effort to catch up to companies like Microsoft and Amazon, the company has been relying on Vertex AI ever since. They are also developing generative AI tools for developers. Developers may use Vertex AI to train models, categorize data, and set up models for production, in addition to constructing generative AI on top of Gemini. It will be interesting whether it moves to expand its walled garden to models beyond those created by Google itself.

Google has been constructing “Chirp” speech services for years, going back to adopting the word as a code name for its early efforts to fight against Amazon’s Alexa service.A