Skoda has created a smartphone app that can listen to your car’s engine and determine if it needs any work completed.
Technicians simply hold the phone close to the engine as it is running, and the sound of the engine is recorded through the microphone. This recording is converted to a spectrogram – a visual depiction of the acoustic signals – then artificial intelligence (AI) compares the recording against stored data to determine discrepancies.
Read more innovative uses for AI and machine learning:
- Artificial intelligence to track UK invasive plant species
- New test uses machine learning to predict premature births
- AI crossword puzzle solver helps machines learn language
Not only can the app tell that something is wrong, it can also understand exactly what is wrong, suggesting servicing measures that might be needed, according to Skoda.
The Sound Analyser app was developed by Skoda’s in-house software developers and currently has a success rate of about 90 per cent. So far it can interpret 10 patterns, including issues with the air conditioning compressor, steering, and clutches in the automatic gearbox.
A total of 245 Skoda dealerships in 14 countries have contributed to the project by sharing recordings for the software’s learning process and trialling the app.
“Sound Analyser is a prime example of the new opportunities digitalisation at Skoda can create, even in terms of after sales," Stanislav Pekař, head of after sales at Skoda, said.
Looking for more smart tech?
- 59 cool gadgets: our pick of the best new tech for 2020
- Best video doorbells: 8 smart doorbells to keep your home safe
- Best wireless earbuds 2020: Budget, noise cancelling and premium
"We will continue to consistently use artificial intelligence technologies to offer our customers an even more personalised service, thus enhancing the customer experience even further.”
The Czech car manufacturer says it is putting artificial intelligence at the core of its digitalisation strategy, using apps such as the Sound Analyser to solve specific problems.
