South China Morning Post
Advertisement
Advertisement
Artificial intelligence
Get more with myNEWS
A personalised news feed of stories that matter to you
Learn more
The Smart Recording Pen A1 is designed to clearly record sound from all directions up to 5 meters (16.4 feet) away. It also doesn’t look like a pen. (Picture: iFlytek)
AbacusTech

This voice recorder looks like a cassette tape and uses AI to transcribe 10 languages

The iFlytek Smart Recording Pen A1 can turn an hour’s worth of recordings into text within five minutes

Artificial intelligence
Karen Chiu
Karen Chiu
Why you can trust SCMP
This article originally appeared on ABACUS

If you’re starting to zone out after too many Zoom meetings, you might want to pay attention to the latest gadget from iFlytek.

The Smart Recording Pen A1 is a recorder that transcribes voice to text. It’s said to support 10 languages, including English, Japanese, Korean and Spanish. The company also says it supports 12 Chinese dialects, including Cantonese. It also looks nothing like a pen, though older people tell me it actually resembles something called a cassette tape.
The Smart Recording Pen A1 is designed to clearly record sound from all directions up to 5 meters (16.4 feet) away. It also doesn’t look like a pen. (Picture: iFlytek)

The 399 yuan (US$56) gadget needs five minutes to transcribe each hour of content. It uses Bluetooth to connect to a smartphone app, which you can open to read the transcript. And don’t worry about your bilingual recordings: iFlytek says the Recording Pen can transcribe a person switching between Chinese and English words.

If you’re wondering why you can’t just use your phone to record and then pay to use iFlytek’s online transcription service… well, actually, you can. But iFlytek says its gadget has enough battery power to record 20 hours of uninterrupted speech and can store 32GB of recordings. That could be useful if you’re a secretary constantly going to corporate meetings or a student who wants a record of every lecture in a semester. Or a reporter with more than few interviews to transcribe.

To test iFlytek’s accuracy, I uploaded a three-minute English audio recording to its website using a free trial. It appears to live up to its 98% accuracy claim, missing only specific terms such as “Minecraft” and some words that were spoken from a distance and, to be fair, weren’t very clear. iFlytek says its algorithm can also translate foreign language transcripts into Chinese, but we weren’t able to test this feature without paying.
The recorder weighs just 30g (1.06oz), which is lighter than a pack of two Reese’s Peanut Butter Cups or the AirPods case. (Picture: iFlytek)

More than any high-tech features, though, it was the cassette-tape look that brought back memories for nostalgic users old enough to remember a time before smartphones and AI.

“When I was a kid, my buddies would record our favorite songs on cassette tapes and give them out to each other,” one Weibo user commented on iFlytek’s post. “We recorded classical music, pop songs and horror stories told on radio. All those wonderful sounds were delivered through heaps and heaps of tapes.”

“After graduation I went to Hunan and bought a Sony Walkman,” another wrote. “I spent a crazy amount of time listening to songs, spending most of my salary on tapes. I took great care of even the printed lyrics sheet inside the box. Whenever the tape was stuck, I would stick a pencil through the reel to unwind it.”

iFlytek is one of China’s leading artificial intelligence companies. Along with facial recognition startups SenseTime, Megvii and Yitu, it was added to a US trade blacklist last year over its alleged involvement in Beijing’s mass detainment of Muslim Uygurs in China’s western Xinjiang province.
iFlytek specializes in AI voice recognition, though it has been previously accused of relying on human interpreters to execute real-time translation at a conference.
Post