| Mobile Web

Microsoft launches in-house voice and image AI models with aggressive pricing

Microsoft has unveiled three in-house AI models covering speech transcription, voice generation and image creation, a report said. The models are available through Microsoft Foundry and MAI Playground. The company said its transcription model posted a 3.8 percent average word error rate across the top 25 languages on the FLEURS benchmark and outperformed rival systems in most languages. It is also applying aggressive pricing and testing the model in Copilot and Teams features.