Saudi voice AI startup Nabrah has launched a real-time speech-to-text service that allows developers and companies to process speech as it is being spoken and integrate the capability into their applications.
The streaming service converts speech without waiting for a recording to finish. Nabrah said its response time is as low as 160 milliseconds, supporting use cases such as voice agents, live call and meeting transcription, and captions for live broadcasts.
The service supports Arabic and English, including code-switching between the two languages within the same conversation. It also includes a word-boosting feature that lets developers prioritise product names, brands and specialised terms to improve recognition accuracy.
Nabrah is offering the service through an API, enabling developers to connect it to conversational applications, customer service platforms, meeting tools and other voice-enabled products. The company said the API can also pass the generated text to other systems for analysis, request understanding or automated actions.
Hosted locally in Saudi Arabia, the service is aimed in part at businesses and developers with requirements around data hosting and local infrastructure. Nabrah said the technology is available for testing and integration through its website.
Source: Jawlah
