Microsoft Opens Custom Neural Voice Platform for Brand Synthetic Speech
Microsoft launches Custom Neural Voice, a service that lets companies create highly realistic synthetic voices using AI. The platform includes strict ethical guidelines to prevent misuse while enabling brands to build custom voice actors for customer service and marketing.
Microsoft unveils Custom Neural Voice, a new service within Azure Cognitive Services that allows companies to create their own artificial voice actors. Brands like Warner Bros. and Progressive already use this technology, but Microsoft now accepts applications from any business interested in developing a custom synthetic voice. The system relies on advanced machine learning models to replicate the unique prosody, pitch, and tone of a human speaker.
The platform works by having voice actors record specific scripts, which the AI then analyzes to predict how that person would sound when saying entirely new phrases. By projecting these vocal characteristics during generation, the synthetic audio achieves a highly natural and realistic quality. Companies currently leverage these custom voices for a variety of applications, including interactive characters like Bugs Bunny at AT&T stores, a chatbot version of Progressive's spokesperson Flo, and a full cast of characters for the Duolingo language app.
To prevent potential abuse, Microsoft maintains strict approval rights over the creation of these synthetic voices. Voice actors must explicitly consent to the use of their vocal likeness, and customers are required to follow a strict code of conduct that includes disclosing when a voice is synthetic. Microsoft also plans to embed digital markers directly into the audio files in the future to ensure there is never any confusion about whether a voice belongs to a human or an AI.