Text to speech is a feature of Azure Speech in Foundry Tools (formerly AI Speech), Microsoft's cloud speech service. It converts text into synthesized speech, a process also called speech synthesis. Its standard voices are neural voices, generated by deep neural networks, in more than 100 languages and locales. A business can also create a custom voice for its brand from a set of audio files and their transcripts; access to custom voice is limited for responsible use.
A program sends text through the Speech SDK or the REST API and receives audio in real time; a batch synthesis API produces audio files longer than 10 minutes, such as audiobooks. SSML (Speech Synthesis Markup Language, an XML format) controls pitch, pauses, pronunciation, speaking rate, volume and speaking style. Viseme events return lip and jaw positions that can animate a face. Each standard voice is available at 24 kHz and 48 kHz. Speech Studio offers a no-code tool for creating audio.
Use is billed per character of input text, with no upfront cost, and a free tier exists.