Amazon is the latest technical giant that unveils a voice -ai model. According to Amazon, the Nova Sonic ‘is a new foundation model that unites speech concept and speech generation in a single model to enable more human voice conversations in AI applications. ” Nova Sonic will compete with similar AI models through Openai, Google and other technical companies.

Nova Sonic understands more than words

Not only does the Nova Sonic understand the speaker’s words, but it can also process the tone, style and pace. The AI ​​voter generator adapts to the conversational context, so that dialogue flows more naturally, compared to the more styled models from the first generations of Alexa. The Nova Sonic can do this because it combines multiple speech processing and features in a single AI model instead of using several different models.

Traditionally, AI Voice Tools involved that several models are performed in order: a speech recognition model would convert speech to text, then a major language model (LLM) would process the import text and generate answers, and eventually a text-to-speech model would convert text to sound. This complex pipeline often removed the tone, style and rate of the original speaker dialogue.

Since the Nova Sonic combines it all in one model, it can adapt to the acoustic context of the input speech. It also responds more naturally to the cadence of human speech; For example, it will not interrupt if the speaker hesitates or stands still to breathe.

How to get Nova Sonic

Nova Sonic is currently available via a new API in Amazon Bedrock, the company’s business for enterprise applications, and will simplify the development of voice applications.

What developers should know about Amazon Nova

The technical giant recently launched Amazon Nova Act, a new AI model trained to perform actions within a web browser. In addition, there is an Amazon Nova SDK for developers to explore. One of the foundation models is Nova cloth to generate high quality images; There are also models to generate text from different modalities, as well as videos from text and image imports.