aiOla Unveils AI Model That Instantly Adapts to Industry Jargon
aiOla, a speech recognition technology provider, has introduced an artificial intelligence model that can instantly adapt to the unique vocabulary of any industry without re-training.
aiOla is enabling companies to capture valuable, previously uncaptured data with speech recognition technology specifically tailored to their business jargon and needs. aiOla's technology supports more than 100 languages and transcribes even heavily accented speech.
Through a model architecture that leverages prompt guidance, it effectively incorporates domain-specific jargon, enabling AI speech recognition systems with zero retraining.
"Enterprises across every industry are acutely aware of the pressing need to adopt AI to maintain a competitive edge, but they don't know where to begin," said Mitch Garber, executive chairman of aiOla, in a statement. "While text-based AI solutions are great for office environments, speech interfaces reign supreme for industrial settings because they seamlessly integrate into existing workflows and collect previously uncaptured spoken data. Prior AI speech recognition models couldn't perform for business use cases because of their inability to grasp jargon. Today, aiOla is changing that by providing instantly tailored AI models that can understand the unique jargon of your specific industry, your organization, or even your team."
aiOla uses a two-step process: initially, the presence of specific terms is detected through aiOla's advanced keyword spotting model, and this information is then used by the company's Whisper-based model to augment its overall speech recognition capabilities and correctly detect the jargon words or terms.
"For this use case, we decided to enhance the most accurate speech recognition model out there, OpenAI's Whisper," said Gil Hetz, vice president of research at aiOla, in a statement. "However, you can apply this approach to any speech recognition model, including Meta's MMS model and proprietary models, unlocking further potential to elevate even the highest-performing speech-to-text models."
aiOla's model's capability to understand jargon instantaneously is achieved by initially freezing the main speech recognition model and adding a proprietary adaptive layer. This adapter undergoes one-time training to guide it in using a vocabulary of jargon, while the model's core capabilities of general-purpose speech recognition are retained. After training, the jargon vocabulary can be hot-swapped to the jargon of a different sector, achieving state-of-the-art performance in recognizing both industry-specific language and general speech.
Related Articles
aiOla's open-source Whisper-Medusa model improves on OpenAI's Whisper by using multi-head attention to speed up and reduce latency while retaining full speech recognition accuracy.
05 Aug 2024