At the forefront of real-time multilingual transcription, Meta Superintelligence Labs has introduced Muse Voice Transcribe, promising significant advancements in speech-to-text technology. This innovative model is designed for developers focusing on real-time transcription tasks in diverse applications, from voice agents and call intelligence to live transcription and high-volume transcription products. With a pricing model of USD 0.18 per hour, accessible via the Meta Model API, Muse Voice Transcribe offers a competitive edge in the growing transcription market.

Trained on 70 languages, Muse Voice Transcribe initially validates 25 fully to ensure high accuracy and reliability. This level of linguistic coverage places Muse firmly among the leaders in the transcription field, highlighted by its top ranking in Artificial Analysis' evaluation for streaming speech-to-text. The model stands out with its ability to dynamically adjust to context, providing an optimal balance between transcription accuracy and latency for each word. Furthermore, Muse’s design allows it to handle code-switching seamlessly, both within and between sentences, enhancing its utility across multilingual environments — features that, as Slator suggests, mark a significant step forward for Meta in the transcription realm.

Functionality-wise, Muse Voice Transcribe excels in identifying up to 20 different speakers and can process audio lengths of up to one hour without requiring additional post-processing. It also uniquely returns timestamps for speaker turns rather than individual words, offering a practical approach for applications needing speaker differentiation. This capability is critical in sectors that rely heavily on nuanced transcription such as legal or clinical settings, where the clarity of speaker attribution can be crucial.

The launch of Muse Voice Transcribe on September 1, 2026, positions Meta alongside other major AI developers in the transcription sector, anticipating upcoming releases like Google’s Gemini 3.5 Transcribe and OpenAI’s GPT-Live-Transcribe. This entry into an already competitive landscape not only underscores Muse's technological prowess but also highlights Meta's commitment to innovation in real-time multilingual transcription services. With Omnilingual ASR’s earlier success, Meta continues to solidify its standing as a leader in the development of sophisticated, open-weight transcription models, setting the stage for further advancements in the ever-evolving landscape of language technology.