Features & Roadmap

Transcribe provides attainable APIs for agency systems to incorporate speech-to-text services into their existing platforms.

Hosted on the Government on Commercial Cloud (GCC) and supports up to RESTRICTED \ SENSITIVE NORMAL data.

Users can upload a pre-recorded audio or video file for processing. A transcript of audio received by a laptop microphone can also be attained instantaneously.

Coverage includes four languages (English, Chinese, Malay, and Tamil), multiple audio formats (e.g. mp3, wav, m4a, wma, aac) and video formats (e.g. mp4, avi).

Google & I2R model in CY engine.

A workspace can be created with the option of adding other Transcribe users, offering controlled access.

Multiple users can enjoy access to the transcript during the live session, instead of waiting for its completion.

Text generated will be matched word-by-word with the narration from the audio files.

• Expand speech-recognition capabilities to capture, analyse, and leverage additional facets of complex human speech, to derive even greater insights.
• Avail self-hosted open-source LLMs to handle SENSITIVE HIGH and SECRET use cases.
• Ability to support data up to CONFIDENTIAL (CLOUD-ELIGIBLE) \ SENSITIVE HIGH.
• Implement more cost-effective and higher performance scaling strategies.

• Improve UI to more intuitively access important information and insights (enhancements to summary, minutes, dashboard, and transcript views).
• Implement RAG on Workspace (via working with GovText) to facilitate information retrieval across a large set of transcripts to easily find answers from past data.
• Improve access to Transcribe Workspace by working with Workpal mobile app, Teams Graph API, SNET, etc.