Amazon has launched a brand-new AI voice model, Nova Sonic, which is set to disrupt the AI voice market with its high quality at a low price.

Amazon has launched a new generation of generative AI model, Nova Sonic, which can natively process speech and generate natural and fluent voice output. Nova Sonic is provided through the Bedrock platform and is priced approximately 80% lower than OpenAI’s GPT-4o. It demonstrates outstanding performance in areas such as speech recognition error rate, multi-language support, conversation quality, and response speed, with an average word error rate of only 4.2% and a response time of 1.09 seconds, making it faster than the GPT-4o model.