Vox Dictum, on-device transcription with speaker diarisation and AI summaries (cobaltinfx.com)

Cobalt InFX has launched Vox Dictum, a privacy-centric transcription software for macOS that operates entirely on-device, ensuring user data remains secure and confidential. With capabilities such as high-accuracy speech recognition for over 60 languages, automatic speaker labeling, and AI-generated summaries, Vox Dictum is designed for professionals who prioritize precision in meetings, interviews, and podcasts. Users can easily import audio or video files, customize speaker names, and generate structured summaries, all without any cloud processing or data sharing. This development is significant for the AI and machine learning community as it showcases the potential of on-device processing and local neural network inference, eliminating traditional concerns about user privacy and data security. By leveraging on-device resources, Vox Dictum not only enhances user control but also mitigates latency in processing, facilitating real-time transcription and summarization. Additionally, the app features advanced capabilities like background noise removal and overlap detection, further improving audio fidelity for clearer outputs. The subscription model integrates seamlessly through the Mac App Store, giving users a straightforward path to access this innovative transcription tool without compromising their data privacy.