Skoda: Sound Analyser detects problems
ARTIFICIAL INTELLIGENCE With the Sound Analyser, the After Sales and DigiLab departments at Škoda Auto are testing a new smartphone app. Škoda Switzerland has also been supporting the pilot project since the beginning of this year. Sound Analyser supports the process of identifying upcoming maintenance needs quickly and accurately with the help of artificial intelligence. The program records the operating sounds of the respective [...]
The Sound Analyzer supports the process of identifying upcoming maintenance needs quickly and precisely with the help of artificial intelligence. The program records the operating sounds of the respective vehicle and compares them with stored sound patterns.
If discrepancies occur, the app uses an algorithm to determine what might be causing these discrepancies and how they can be resolved.
Diagnostics of important wear parts
The Sound Analyser app uses artificial intelligence technologies to reliably, unambiguously, and quickly assess the current condition of wear parts and indicate necessary maintenance work.
To do this, the program includes various vehicle-specific parameters and analyzes the usage profile of the respective vehicle. The Sound Analyzer thus facilitates precise diagnostics for service technicians, who only need a standard smartphone or tablet to use the app.
Prerequisite are concrete descriptions
The use of the technically highly complex program is intuitive: The operating noise of the vehicle can be recorded via the smartphone using the app. In the next step, an algorithm compares the recording with stored sound patterns and provides a concrete description of the result on this basis.
Currently, the app already recognizes ten patterns with an accuracy of over 90 percent, including assemblies such as the steering gear, the air-conditioning compressor and the clutches of the direct-shift gearbox (DSG).
At the same time, the app is conditioned to recognize other sound patterns. You can see application examples in the gallery.
Representation through artificial intelligence
The Sound Analyzer uses neural network algorithms as its technical basis. The app converts the sound recording into a spectrogram that depicts sound signals pictorially.
Artificial intelligence is then used to compare the display with the stored values in order to identify deviations. On this basis, the app classifies any maintenance or repair requirements according to defined models.
Product support by Amag Import AG
The smartphone app has been running in trial operation in 14 countries since June 2019, including the markets of Germany, Russia, Austria or France.
Since the beginning of 2020, Switzerland has also been involved in the pilot project with 18 dealers and the product support of Amag Import AG. A total of 245 dealers are participating. They provide the sound recordings for the software learning process and thus contribute to the further development of the program.
50 to 100 shots are needed
Eric Dénervaud, Product Supporter at Škoda Switzerland: "For the app to recognize and classify other sounds, it needs at least between 50 and 100 recordings of the same sound. The more sounds the app can identify and distinguish, the more efficient it becomes."
Pillar of the current digitization strategy
Artificial intelligence is a pillar of Škoda Auto's digitization strategy. On the basis of artificial intelligence, technologies perform perceptible functions that are otherwise the exclusive domain of humans.
In this way, corresponding programs can interact with the environment, perceive and weigh facts or solve certain problems.
