Skip to main content

Baidu’s SwiftScribe uses AI to transcribe audio files up to an hour in length

baidu
Image used with permission by copyright holder
Baidu may be known as “the Google of China,” but that doesn’t mean the Asian search giant doesn’t come up with its own unique applications. On Monday, it debuted SwiftScribe, a web app that automatically transcribes speech files with the help of artificial intelligence.

SwiftScribe is about as simple as web apps come. It recognizes files in .wav and .mp3 format, and once the upload’s complete, the transcription process gets underway. A 30-second file takes about 10 seconds, and a one-minute file less than 30. An hour of audio, the maximum length SwiftScribe will allow, takes 20 minutes.

Recommended Videos

It’s not always perfect. SwiftScribe sometimes misses the spelling of certain words, and capitalization and punctuation aren’t always on point. But it offers an editable field that lets users correct mistakes, and a built-in speed-shifting tool that plays the uploaded audio clip audio at a faster or slower speed.

Please enable Javascript to view this content

Baidu project manager Tian Wu, who was inspired partly by her experience transcribing interviews as a graduate student at the University of California, Santa Barbara, said that SwiftScribe has the potential to save hours. “English is not my first language,” Wu told VentureBeat. “It took 10 hours to transcribe one hour of audio. That’s my personal experience. Usually, it will take a professional four to six hours to transcribe a one-hour audio clip.”

Image used with permission by copyright holder

Wu told VentureBeat that SwiftScribe can help transcribe audio 1.67 times faster on average. She envisions transcriptionists doing more work and ultimately getting paid more for it.

SwiftScribe’s more proof of concept than polished product, right now. In the coming months, the team plans to enhance the app with video transcription and captioning, support for more file formats, and an option for automatically adding punctuation.

It’s free to use for now, but Baidu’s considering a paid option. “In the future, we hope to turn it into a business,” Wu said.

Baidu may not have the name recognition in the United States that it does in mainland China, where the Beijing-based juggernaut commands roughly 80 percent of the internet search market and amasses quarterly profits that regularly top the hundreds of millions. But it’s hoping to change that. In 2013, it opened the Institute of Deep Learning, a research center devoted to advancing the firm’s artificial intelligence efforts.

In the immediate future, the Chinese aims to use the lab to increase revenue by building augmented reality marketing tools. But it may be considering a significant expansion of health-care and education applications.

Kyle Wiggers
Former Digital Trends Contributor
Kyle Wiggers is a writer, Web designer, and podcaster with an acute interest in all things tech. When not reviewing gadgets…
Sundar Pichai says even more AI is coming to Google Search in 2025
Google Search on a laptop

Google will continue to go all in on AI in 2025, CEO Sundar Pichai announced during the company's Q4 earnings call Wednesday. Alphabet shares have since dropped more than 7% on news that the company giant fell short of fourth-quarter revenue expectations and announced an ambitious spending plan for its AI development.

"As AI continues to expand the universe of queries that people can ask, 2025 is going to be one of the biggest years for search innovation yet,” he said during the call. Pichai added that Search is on a “journey” from simply presenting a list of links to offering a more Assistant-like experience. Whether users actually want that, remains to be seen.

Read more
Apple’s AI-focused M5 chip enters mass production
MacBook Pro with M4

Apple has begun the mass production of its M5 chip, which is set to power next-generation products, including the upcoming Mac series and iPad. Coinciding with long-standing reports, the Cupertino-based tech brand is establishing a new node process for packaging the semiconductor. The technology is intended to provide improved AI performance on the devices it powers, according to ETnews.

Industry sources told the Korean publication that Apple began packaging the M5 chip last month. Taiwan's TSMC began the initial production of the M5 chip circuit using its 3nm process (N3P). The technology is expected to improve the power efficiency of the M5 chip by between 5% and 10%, and performance by 5% in comparison to the previous M4 chip, which will aid in improving AI performance on upcoming Mac and iPad models.

Read more
Google says quantum computing applications are five years away
Google Quantum chip Willow.

A few weeks ago at CES 2025, Nvidia CEO Jensen Huang posited that practical uses of quantum computing were about 20 years away. Today, Google’s head of quantum Hartmut Neven told Reuters that we could see real-world applications of quantum computing within five years. So, who is right?

According to Huang, current quantum systems don’t have enough “qubits.” In fact, they’re short by around five or six orders of magnitude. But why do we need so many? Well, current research suggests that more qubits result in fewer errors, creating more accurate quantum computers. Let's talk about why that is.

Read more