Microsoft released MAI-Transcribe-2, an AI speech-to-text model that achieves 10x faster performance and improved accuracy (5.2% error rate vs. competitors, 3.4% for Thai) with multi-speaker diarization, word-level timestamps, and 60-language support. The service is priced at $0.10/hour with special pricing through end of year, currently available via Microsoft Foundry, MAI Playground, and Open Router—a significant upgrade for Thai and multilingual content creators and developers.
← Back to all articles