AppTek

AppTek is a provider of speech and language technology built on artificial intelligence and machine learning. Its engines cover automatic speech recognition, neural machine translation, natural language processing and understanding, large language models and text-to-speech. AppTek serves organizations in media and entertainment, call centers, government and enterprise business, and offers its engines to enterprise customers and partners who integrate them into their own content and applications.

AppTek delivers its technology as a cloud API with a web portal or as an on-premise service, and processes audio in streaming live or batch offline modes. Customer projects follow three steps: discovery and learning, training and testing AI models, and application deployment. AppTek's scientists build custom models for a customer's proprietary content or subject domain, including new language models for low-resource languages reserved for that customer's exclusive use.

The speech-to-text engine adds punctuation, capitalization and number formatting, separates speakers through diarization or separate channels, indexes timestamps against words, handles custom lexicons, adapts to noisy audio and can redact numeric data such as credit card numbers. The translation engine offers pre-trained models in hundreds of language pairs, document translation, website localization, live media captions, language detection and custom terminology. Neural text-to-speech produces synthesized voices, including custom voices built from a customer's recordings.

AppTek lists speech recognition in more than 60 languages and dialects for narrowband telephony and wideband broadcast audio, including Arabic in 13 dialects and English in 14 dialects. Live demos are available for the base speech recognition and translation models, and a Speech Translate mobile application can be downloaded. AppTek also develops ClimateGPT, an open source large language model family for climate science. Pricing is arranged directly with AppTek through its contact form or a scheduled demo.

Use cases

• Transcribing contact center conversations, voice commands and other spoken content
• Creating real-time closed captions for live media and archiving media assets
• Speech-to-text assistive technology for hard-of-hearing people
• Transcribing all conversations to check compliance with industry regulations
• Transcribing witness and subject statements for keyword search
• Speech analytics to gauge customer sentiment and brand perception
• Website localization, document translation and multilingual broadcast news captioning