A new, enterprise-specific AI speech model is here: Jargonic from aiOla claims to best rivals at your business’s lingo

Speech recognition forms have become increasingly accurate in recent years. However, they may be built and measured under perfect circumstances-oily rooms, vocabulary and public vocabulary. For institutions, however, the sound in the real world is more chaotic.

This is the challenge that AIOLA aims to address through the launch of Jargonic, designed to identify new automatic speech (ASR) specifically to use the institution. Israeli youth reveals Jargonic today.

Jargonic is a new model of speech to a designed text to deal with specialized terms, background noise and various accents without or adjusting wide -ranging.

“Our model focuses on three main challenges in recognizing speech: terms, background noise and dialects,” said Gil Hitz, Vice president AIION AI. “We have built a model that understands the defined industrial terms in a zero way, deals with loud environments and supports a wide range of dialects.”

Avi is now available on the AIOLA Foundation platform, Jargonic as ASR is ready to produce companies in industries such as manufacturing, logistical services, financial services and health care.

From the first product to the first intelligence

Jargonic’s launch is a shift in the focus on Aiola itself. According to the leadership of the company, the team redefined its approach to setting the priorities of artificial intelligence and publishing research.

“When I got here, I saw an amazing product company that has invested extensively in the capabilities of advanced artificial intelligence, but I was mostly famous for helping people fill out models,” said Assaf Asbag, chief technology and products official in Ayla. “We have turned the perspective and we have become Amnesty International with a great product, instead of AI’s products.”

“We decided to open our capabilities to the world,” added ASBAG. “Instead of serving our model only for institutions within our products, we have developed an application programming interface and now launches it to make our resistant model available to all.”

Learn about terms, zero adaptation

One of the distinctive Jargonic features is its approach to specialized vocabulary. Speech recognition systems usually fight when facing the term terms that do not appear in standard training data. Jargonic deals this challenge through the special keyword discovery system that allows zero adaptation-Participants can simply submit a list of conditions without additional re-training.

In standard tests, Jargonica showed 5.91 % average word error rate (WER) Through four leading academic data collections, it surpasses competitors such as Eleven Labs, Assembly Ai, Openai’s Whisper and DeepGRAM NOVA-3.

However, the company did not reveal comparisons of performance specifically against the most newest multimedia models such as Openai GPT-4O TRECRIBE, which came nine days ago, characterized by better performance on standards like Wer, with only 2.46 % in English. Aiola claims that its model is still better in choosing Jargon specific works.

Jargonic also achieved 89.3 % of the recall rate According to specialized financial conditions, it continuously outperformed others in identifying multi -language terms, and reaching more 95 % accuracy Through five languages.

“Once you get heavy terms, the accuracy of recognition usually decreases by 20 %,” ASBAG explained. “But through our zero approach, where you can only include important keywords, the accuracy jumps to 95 %. This is unique for us.”

This possibility is designed to eliminate a long -time re -training process, and it is usually required to adapt ASR systems with specific industries.

It has been improved for the institution’s environment

Jargonic Development has been informed of years of experience in building solutions to institutional customers. The model has been trained in more than a million hours of written speech, including important data from industrial and commercial environments, ensuring durability in loud realistic environments.

“What distinguishes us is that we have spent years solving the problems of institutions in the real world,” Hitz said. “We have improved speed, accuracy and the ability to deal with complex environments-not only podcasts or videos, but noisy, chaotic, and realistic workplaces.”

The structure of the model combines the keyword that is discovered directly into the copying process, allowing Jargonic to maintain accuracy even in unpredictable sound conditions.

The first future

For the leadership of Ayla, Jargonic is a step towards a broader shift in how people interact with technology. The company does not see recognition of speech not only as a work tool, but as a basic interface for the future of human and computer interaction.

“Our vision is that every facade of a machine will soon be the first voice,” Hitz said. “You will be able to talk to your refrigerator, emptiness, and any machine – it will act and do what you want. This is the future that we build towards.”

ASBAG repeated these feelings, adding, “AI will become a new web browser. The machines have started to understand, and now we have a reason to interact with them normally.”

At the present time, the focus of the institution remains on the institution. Jargonic is immediately available to API customers, allowing them to integrate the possibilities of speech recognition in the form of work, applications, or services facing customers.