TranslateLive and Descript facilitate communication with language-processing AI.

Of all the types of AI currently in use, none has become so ubiquitous – and yet remains so invisible – as natural language processing (NLP).

From chatbots that automatically respond to customer service queries to virtual assistants that recognize and obey voice commands, the field dedicated to enabling computers to work with human language has greatly advanced in just the last decade. But, as several innovations being rolled out demonstrate, there is still much room to grow.

TranslateLive and Descript are two companies pushing the boundaries of distinct but complementary facets of NLP: one uses it to overcome language barriers (translation), while the other goes beyond the confines of oral communication (transcription). 

Overcoming language barriers

Based in Florida, TranslateLive allows two people who do not speak the same language to communicate via real-time speech-to-text conversion and translation without having to download a separate application.

The TranslateLive platform uses Compute Engine, Translation API, and Cloud Speech-to-Text to drive its live translation features.

Compute Engine provides the necessary performance and scalability for real-time translation without latency issues, while Translation API and Cloud Speech-to-Text handle additional translation needs within the TranslateLive browser tool and mobile application.

Google language intelligence also helps TranslateLive understand and accurately translate different accents, while Google Cloud Platform infrastructure lets TranslateLive focus on overall development instead of small, but time-consuming tasks such as setting up local servers.

Real-time, in-app translation on the ILA device
Source: TranslateLive

Going beyond the spoken word

Descript offers an AI-enabled application that syncs audio to automated transcripts and provides easy editing capabilities. From legal proceedings to entertainment content, quickly and accurately transcribing audio materials is a time-consuming job.

Many automated programs exist, but they tend to be limited by the quality of the original audio source; from distracting background noises to mumbling speakers, there are many factors affecting audio clarity and hence the quality and accuracy of the transcription. To provide the highest possible level of word accuracy, Descript uses the REV engine to recognize speech and edit transcriptions in real time.

Fast and accurate transcription of audio files

Enabling cooperation across cultures and territories

The services offered by both TranslateLive and Descript provide immediate and tangible benefits to a wide range of users and organizations.

TranslateLive is a cost-effective platform for non-profits and other non-governmental organizations to enhance internal and external communications in different languages. Call-center agents also use the service to speak with customers in different languages, preventing any potential miscommunications or misunderstandings.

TranslateLive’s sister company VTCSecure also works in the deaf communications space and is exploring further applications with accessibility-friendly benefits. 

Meanwhile, Descript limits costs, improves workflows, and reduces error margins for a range of content producers who need fast and accurate transcriptions (journalists, radio stations, legal departments). The company is also testing context-aware recognition to improve contextual speech recognition, providing specific jargon glossaries for different industries.

Beyond these applications, both TranslateLive and Descript are redefining the way people create and interact with voice- and audio-based media – pushing the boundaries of cross-cultural, cross-border, and cross-platform communications for the long-term betterment of societies and economies.