Journalist-built speech-to-text startup enlists same tech as UN humanitarian efforts
Speech-to-text startup Trint utilizes the same technology as the United Nations Office for the Coordination of Humanitarian Affairs (UN OCHA) as both organizations look to improve the quality of transcription.
Co-Founded by Emmy-winning network television news and war correspondent Jeff Kofman, the DNA of Trint is intrinsically woven in journalism — a profession that relies heavily on transcription.
Kofman estimates he has manually transcribed thousands of hours of interviews, speeches, lectures, and news conferences through his career, and this is an extremely time-consuming task as this author can relate.
Always the journalist, Kofman came up with the name of Trint as a portmanteau of “Transcription” and “Interview,” and the platform is built on some of the same technology that the UN is currently working with to mobilize humanitarian aid for crises such as famines, epidemics, and war.
According to Simon Turvey, VP of Engineering at Trint, “The most exciting thing at the moment is that we are leveraging the very cutting-edge tools and frameworks and languages that are enabling us to execute extremely quickly on our innovation roadmap.”
“We’re heavily leveraging ReactJS to build out next generation of web applications. We’re using tools like Kubernetes on the backend for our scalable infrastructure, and we’re embracing philosophies like infrastructure as code to facilitate devops and make engineering more of a pleasure in terms of getting features deployed and out into people’s hands,” added Turvey.
UN Humanitarian Efforts
On the UN front, a social data ingestion, analysis, and visualization platform called the Fortis Project partnered with Microsoft to develop a speech-to-text software that would provide “access to tremendous amounts of data and intelligence, from a wide range of sources including newspapers, social media, and radio,” according to the Microsoft Developer Blog.
Software Engineer and Microsoft Developer Blog author, Clemens Wolff, eloquently explained the necessity of speech-to-text software as it relates to the UN mission.
“In many parts of the world, essential information is broadcast via traditional sources such as radio. To formulate effective crisis response plans, the United Nations Office for the Coordination of Humanitarian Affairs needs to be able to access and analyze these data sources in a timely manner.”
Just as Microsoft teamed up with the UN to improve speech-to-text for clearer reporting, Trint recently joined global community program SPROCKIT, which “curates, connects, and fosters collaboration among leading media, entertainment and technology companies and market-ready startups to bring innovative products, services and revenue models to market.”
SPROCKIT Founder and CEO Harry Glazer remarked, “I’m so pleased to welcome Trint to this year’s SPROCKIT class. Members recognize the media, entertainment and technology sectors are exponentially evolving, and we must work together and learn from each other to quickly adapt and stay innovative.”
Speech-to-text Future of Augmented Journalism
As the media takes a huge hit as a result of digital disruption and transformation, the new era of news production is becoming automated where “algorithms can help break stories and computers have bylines” in what is called Augmented Journalism, according to MediaShift.
In Augmented Journalism “AI capabilities, like voice recognition and speech-to-text conversion, will allow for hands-free work on other more nuanced tasks. With these more menial responsibilities out of the picture, journalists will be freer to take on the big news of the day,” wrote Mayo Nissen and Graham Tuttle on MediaShift.
This “hands-free work” aspect of speech-to-text that will clear up time for those taking advantage of the software is already well underway in not just journalism, as evidenced by the Microsoft and UN partnership.
Likewise, even though Trint was founded by a journalist, its applications are applicable across industries and professions including social media marketers, students, and researchers.
The software can be used for anyone looking to transcribe “an audio recording or a video of two or more speakers (or just one) engaged in natural speech,” reported Wired.
While Trint can distinguish 13 languages and a handful of English dialects, just don’t expect it to accurately transcribe the mumbling of the late gonzo journalist, Hunter S.Thompson; Keith Richards is doable, though.