Open AI Unveils Text-to-Speech Website

Open AI has introduced new speech-to-text and text-to-speech models available via APIs, enabling developers to create personalized voice agents for customer service, transcription, and creative projects.

Tech Mar 22, 2025 0 53 Add to Reading List

Open AI is actively developing the intelligence and functionality of text agents — systems that autonomously perform tasks for users. Key releases include Operator, Deep Research, Computer-Using Agents, and the Responses API with integrated tools. However, for such agents to be truly useful, users must be able to communicate with them naturally, using conversational language.

The company announced the launch of new speech-to-text and text-to-speech models in its API. These models set a new standard for accuracy and reliability, especially in challenging conditions such as noisy environments or accented speech. They are ideal for use in call centers, meeting transcriptions, and other scenarios where accuracy is important.

For the first time, developers will be able to customize voice agents by setting the tone and style of speech, such as “speak like a helpful customer service agent.” Text-to-speech conversion is already available for testing on the openai.fm website .

Since 2022, the company has been actively developing audio models, improving their intelligence and accuracy. New models allow developers to create more reliable and personalized voice solutions, expanding the possibilities of interaction with users through natural spoken language.