MiMo-V2.5 Voice

The software provides automatic speech recognition that operates in two languages simultaneously, handling regional dialects, code‑switching between languages, and musical vocalizations. It is aimed at developers who need to transcribe bilingual audio streams, multilingual dialogues, or sung content where speakers shift between languages or use non‑standard speech patterns. The system is offered as part of the Xiaomi MiMi API Open Platform, where a one‑time purchase grants access to both the MiMo‑V2‑Pro and Omni flagship models, while a text‑synthesis model is available for free for a limited period. Currently classified as experimental, it targets use cases that require flexible, multilingual transcription beyond standard monolingual ASR solutions.