After introducing Copilot Vision, a screen analyzing tool with AI capabilities, Microsoft has a new product in testing. Windows Latest tested Copilot Audio Expressions, an AI voice generation tool that can produce a human-like audio output for your script and even take creative control if you wish it that way.

The tool is available to test on the official Copilot Labs website, where you can try the tool. It has two modes, Emotive and Story, with almost a dozen voices and style options. Windows Latest conducted a deep analysis of how the feature works, trying different voice and tone options, and the tool is quite good at what it does. But it only offers English as the primary language.

Firstly, we started off with a dummy script that narrates a scenario of a busy train station. We fed the script into the prompt field, added a line of script to specify the voice style.

We chose Emote as the mode option with the Oak voice and narration style. The tool didn’t take long to generate an audio, which we were able to play using the integrated web player.

The emote mode generates a maximum of 59 seconds per audio clip. We expected a simplified narration that would be an AI voice reading the text aloud. Surprisingly, the audio clip took the liberty with the supplied script and added bits, rephrased sentences to sound more engaging. It added tiny yet useful details to the already good script snippet.

We like the fact that you can download the audio file without any login requirements, and it uses the MP3 format. So, playing it on any device or player won’t be a problem. Copilot Audio Expressions impressed us with the Emote mode with granular controls, but we wanted to explore the Story mode too.

Generate compelling stories with Story mode

Next, we selected the Story mode that picks the voice and style automatically and offers no option to adjust them. You can specify what you want with the prompt, but we went with a basic prompt of “tell a story about a cat lurking in the shadows for filling prey.”

This time, the tool cooked up a 90-second story about a person narrating a cat looking for prey. It had both the narrator’s voice (in American accent) and the cat’s voice (yes, the cat speaks in this story, expressing its hunger) in a British accent. Both synced up perfectly and sounded like a collaboration rather than a flat AI voice.

You can check out both the audio snippets to check the output quality.

We found it worthy of an AI audio generation tool that you can use to create audio clips. Give the tool a try on the Copilot Labs site.

The post Hands on with Microsoft Copilot’s new audio AI that sounds more personal than ChatGPT appeared first on Windows Latest