DhivehiGPT Releases Next-Generation Text-to-Speech Service With Seven New Voices

DhivehiGPT has released a major upgrade to its text-to-speech service, introducing a completely overhauled TTS engine powered by a brand-new, state-of-the-art model. The release represents a significant advancement for Dhivehi speech technology, delivering the most advanced Dhivehi TTS model currently available on the market.

The upgraded engine produces more realistic and natural-sounding speech, with substantial improvements to pronunciation, rhythm and overall voice quality. It handles numerals, dates, sentence endings and pauses more accurately, allowing generated audio to sound smoother and more expressive across a wider range of content.

The service now includes seven new voices spanning male, female and child voices. Several of the voices were developed using recordings from professional narrators, while others were created in collaboration with voice actors who formally licensed their voices for use within the platform.

In addition to Dhivehi, the new TTS engine supports English and Arabic, enabling users to generate multilingual speech through the same service. This expanded language support makes the platform more versatile for content creators, educators, organizations and developers working with audiences in the Maldives and beyond.

DhivehiGPT remains open to collaborating with additional voice actors who are interested in licensing their voices and contributing to the continued growth of Dhivehi-language speech technology.

The release is the result of extensive research, development and coordination across the Javaabu team. Mohamed Jailam, Co-Founder and CEO of Javaabu, led the effort to source and coordinate the new voices, including collaborations with professional narrators and voice actors.

Arushad Ahmed, Co-Founder and CTO, led the technical development of the upgraded TTS engine. Machine Learning Research Engineer Ahmed Yameen played a key role in improving, scaling and deploying the new model, while Creative Multimedia Designer Mohamed Anaadh Hussain contributed through extensive model research, testing and fine-tuning.

The achievement reflects the team’s technical skill, persistence and commitment to advancing Dhivehi-language artificial intelligence. Their collective work has transformed the service from its initial TTS release in 2024 into a significantly more capable and natural speech-generation platform.

The team plans to continue improving the service and, In Sha Allah, introduce more voices, capabilities and features in future releases.

Users can experience the upgraded text-to-speech service now at dhivehigpt.com/tts.