wordpress blog stats
Connect with us

Hi, what are you looking for?

New deep learning model can predict passwords from keystroke sound with 95% accuracy

The paper notes that these types of attacks can be carried out with off-the-shelf equipment and algorithms.

A team of researchers in the United Kingdom developed a deep learning model that can accurately predict what you are typing based on the sounds from keyboard keystrokes, Bleeping Computer reported on August 5.

According to the research paper, the model has a 95 percent accuracy when keystrokes are recorded with a nearby smartphone microphone and a 93 percent accuracy when keystrokes are recorded over a Zoom call.

Why does this matter: The paper notes that these types of attacks, known as acoustic side-channel attacks, can be carried out with off-the-shelf equipment and algorithms. In the wrong hands, this gives bad actors access to sensitive information. For example, imagine you are working from a cafe, and someone can discern all that you are typing, including your login passwords, credit card details or personal messages, just by recording the sound of your keystrokes sitting at the next table. Moreover, the research points towards another concerning issue: the use of artificial intelligence (AI) to develop cyber-attacks. While earlier, it would have required considerable time and other resources to train such classifiers, advances in AI-based deep learning make the process much faster and easier.

How did the researchers develop this model: 

  • To train their model, researchers pressed 36 keys on a MacBook Pro 16-inch (2021) 25 times each and recorded the sound produced by each press using an iPhone 13 mini placed 17 cm away from the laptop. Notably, the same keyboard has been used in all MacBooks for the past two years and is likely to be used in future MacBooks as well.
  • The researchers then made a visual representation of the differences for each key by producing waveforms and spectrograms.
  • These spectrogram images were used to train CoAtNet, an image classifier.
  • The trained CoANet classifier was able to achieve 95 percent accuracy in identifying keystrokes recorded using a smartphone, 93 percent accuracy for recordings through Zoom, and 91.7 percent accuracy for recordings through Skype.

How to safeguard yourself from such attacks: The paper also listed measures you can take to avoid falling victim to such acoustic side-channel attacks. Some of these measures include:

  • By simply changing the typing style. For instance, by using touch typing, the accuracy of such attacks is reduced dramatically.
  • Using randomised passwords with multiple cases can reduce the accuracy of such attacks because of the use of the shift key.
  • You can play white noise or sounds of fake keystrokes to confuse the classifier.
  • Use two-factor authentication methods, such as biometric, for logins.

Also Read

STAY ON TOP OF TECH NEWS: Our daily newsletter with the top story of the day from MediaNama, delivered to your inbox before 9 AM. Click here to sign up today!


Advertisement. Scroll to continue reading.
Written By

Free Reads


The case dates back to 2019 when Spotify filed a complaint against Apple's anti-steering rules which prevented apps like Spotify from informing their users...


DIP contains information regarding cases detected as misuse of telecom resources and acts as a back-end repository of all requests received on the Sanchar...


Meta's transparency report reveals a sharp rise in GAC orders from India in January 2024, exceeding the overall monthly GAC orders sent in the...

MediaNama’s mission is to help build a digital ecosystem which is open, fair, global and competitive.



NPCI CEO Dilip Asbe recently said that what is not written in regulations is a no-go for fintech entities. But following this advice could...


Notably, Indus Appstore will allow app developers to use third-party billing systems for in-app billing without having to pay any commission to Indus, a...


The existing commission-based model, which companies like Uber and Ola have used for a long time and still stick to, has received criticism from...


Factors like Indus not charging developers any commission for in-app payments and antitrust orders issued by India's competition regulator against Google could contribute to...


Is open-sourcing of AI, and the use cases that come with it, a good starting point to discuss the responsibility and liability of AI?...

You May Also Like


Google has released a Google Travel Trends Report which states that branded budget hotel search queries grew 179% year over year (YOY) in India, in...


135 job openings in over 60 companies are listed at our free Digital and Mobile Job Board: If you’re looking for a job, or...


By Aroon Deep and Aditya Chunduru You’re reading it here first: Twitter has complied with government requests to censor 52 tweets that mostly criticised...


Rajesh Kumar* doesn’t have many enemies in life. But, Uber, for which he drives a cab everyday, is starting to look like one, he...

MediaNama is the premier source of information and analysis on Technology Policy in India. More about MediaNama, and contact information, here.

© 2008-2021 Mixed Bag Media Pvt. Ltd. Developed By PixelVJ

Subscribe to our daily newsletter
Your email address:*
Please enter all required fields Click to hide
Correct invalid entries Click to hide

© 2008-2021 Mixed Bag Media Pvt. Ltd. Developed By PixelVJ