Copilot Adds Audio Generation with Expressive Voices

Microsoft Copilot will now be able to create audio natively thanks to another new artificial intelligence (AI) innovation from Microsoft recently added. The Redmond-based tech giant revealed on Wednesday that Copilot will be gaining a new audio generation tool that will allow users to give it a screenplay and have it generate an AI voiceover in various genres. None of the modes will sound like standard text-to-speech models because it is native voice production. Interestingly, the company is using its in-house MAI-Voice-1 AI model to enable this feature.

The Copilot Labs experiment called Copilot Audio Expressions converts written language into a conversational, lifelike voice, complete with personality, emotion, and style. Whether you want a soft tone for a bedtime story, a positive message for a motivational speech, or a theatrical flare for an epic adventure, you can change the voice’s tone to fit your prompt. Downloading audio samples for use in content development is simple.

Microsoft AI CEO Mustafa Suleyman announced the launch of Copilot’s new audio generating options in a post on X (previously Twitter). He emphasised that the MAI-Voice-1 AI model, which was made public at the end of August, is what powers these. Only Copilot Labs currently offers this experience when logging in with a personal account.

The quick of this awesome video on X tells a lot about this new feature introduced to the Microsoft Copilot.

You can experiment with the three modes. The first is the scripted mode, in which the AI chatbot reads the input exactly as it is, without additional style or flair. These work best for jobs like information presenting, document narration, and formal announcements of any kind.

Emotive is the name of the second mode. It main goal according to Suleyman, is to make the input seem appealing and dramatic. In order to deliver a performative piece, the voice will use a variety of intonation, pitch, and tone. This is perfect for casual narration, marketing, and advertising.

Story is Copilot’s last mode for audio generation. With many voices and characters, this format is the most adaptable. The mode according to the company, is perfect for activities involving analysis, podcast-style presentations, and narrative. Despite the fact that Microsoft has not specified any rate limits, the feature is now free to use. When the feature will be added to the Copilot desktop and mobile apps is unknown.

Additionally, it Story Mode, also is one which combines several speaking types to create captivating, approachable storytelling experiences that are ideal for parents, teachers, and others. A sneak peek at how AI can enhance audio content’s expressiveness, immersion, and enjoyment for all is provided by Audio Expressions.

Microsoft claimed that the MAI-Voice-1 was a speech generation model that could naturally produce expressive, natural-sounding voices at the time of release. On a single GPU, it can produce a whole minute of audio in less than a second. The model was trained on approximately 15,000 Nvidia GPUs by the tech giant.

Microsoft further says everyone can use Copilot Audio Expressions for free as it is accessible everywhere and compatible with all present browsers. This experimental tool only supports English at this time. Although it can be accessed through mobile browsers, we advise using a desktop computer for optimal results.

High-quality MP3 files of all audio expressions are included, perfect for sharing, embedding, or playing on many devices.

The purpose of Copilot Audio Expressions is personal use in accordance with the Copilot Code of Conduct.

Remember that you should only enter content that you are the owner of or have permission to use. Never enter or request audio that violates someone else’s rights (such as copyright, trademark, or publicity rights), their privacy, or the Microsoft Services Agreement or Copilot Terms in any other way.

The company says that it may restrict or suspend user’s account or Copilot access if they find that copilot users are attempting to enter or create content that is against the Copilot Terms Code of Conduct or that is illegal.