Tech

AI transcription is the worst (workarounds are:)

For years, I’ve been looking for a reliable way to autonomously copy natural speech. I’m a journalist and spend hours transcribing recorded interviews with sources around the world. For now, I’m still paying for a manual transcription service.

Speech recognition is a major challenge for AI developers and is a puzzle that is drawing attention in various industries. This technology has implications that go far beyond source citations. Human-machine interfaces in areas such as robotics, self-driving cars, and personal computing benefit from computers that can accurately interpret natural speech.

Transcription is therefore a kind of technical entry point, a direct market need that helps facilitate the development of technologies that have a wide range of resonances and immeasurable impacts on how machines interact.

“As with almost every market segment, the education, legal, media and entertainment industries needed to move quickly to remote environments,” said Jidas, managing director and president of Sapphire Ventures. “As a result, the need for AI-driven, real-time, accurate transcription services has skyrocketed.”

The problem is that, in addition to accents and dialects, speeches in natural contexts have made the quest for AI-driven transcription strange so far. So what if you have a mature market for technology but don’t have that feature yet?

Now, you spend money on technology development and improvise with free tools.

This is a strategy for an innovative transcription and caption solution called Verbit. It leverages in-house AI-based technology and an army of human supervisors to transform live and recorded video and audio into near-perfect captions and higher education transcriptions. Education, legal, media, and enterprise industries.

“Verbit solves this big problem for companies and organizations in these markets by combining the speed and low cost of automatic speech recognition technology with the accuracy of human transcription,” said a venture company recently Verbit 6000. Das, who led the $ 10,000 Series C, says. Verbit is currently over $ 100 million.

Verbit’s model uses state-of-the-art transcription technology that filters out background noise and echo and recognizes domain-specific terminology and more. The acoustics, language, and contextual data are then thoroughly checked by a Verbit human writer. The writer maintains quality assurance by editing and reviewing the material and incorporating customer-provided notes, guidelines, and more. I’m often pleased when the human writer I work with contains few contextual notes about spelling and transcription usage.

I like this strategy very much. Verbit can take advantage of the great needs of major enterprise players: real-time transcription needs, using core technologies that are good but not yet complete. The hybrid human-machine model allows the company to bring high-quality products to market while continuing to invest in development. Despite the dystopian nightmare of robots stealing jobs, automation is a way to penetrate the enterprise in the foreseeable future. By joining forces with humans rather than expelling them completely.

According to a company statement, Verbit will continue to innovate the capabilities of its data-driven products with this latest investment round, further driving significant growth by increasing the number of languages ​​it supports.

AI transcription is the worst (workarounds are:)

Source link AI transcription is the worst (workarounds are:)

Back to top button