Microsoft said today that it would make improvements to the Read Aloud feature in Edge in the new Chromium-based version of the browser.

“A common theme that we heard from Read Aloud feedback in the current version of Microsoft Edge was that the default speaking voices sounded robotic and unnatural,” Microsoft’s Scott Low explains. “People also told us how time-consuming it was to install different language packs so that they could read text in other languages. We have made this experience better with the arrival of cloud-powered speaking voices in the preview builds of Microsoft Edge.”

The new speaking voices are available now in the most recent Dev and Canary channel releases of the new Microsoft Edge. There are 24 of them across 21 different locales, and they are powered by Microsoft Cognitive Services. According to Microsoft, the voices are split between two types—standard and neural, the latter of which are the most natural-sounding.

You can test this feature by navigating to a webpage, selecting some text, right-clicking it, and choosing “Read aloud selection.” Then, choose one of the voices marked “Neural” by selecting Voice options > Choose a voice in the Read Aloud toolbar that appears. The results are indeed impressive.

Microsoft says that these voices are also available to developers through its JavaScript-based SpeechSynthesis API. I assume there will be an easy way to read aloud entire webpages, PDF files, and EPUB files soon as well.