Automatic detection of noise anomalies with AI
Optimize your product quality with AI-based acoustic quality control and detect anomalies early on in the production process.
Reliable fault detection in production and development
Sounce enables automatic detection of unwanted noise in real time, for example for fault detection during the assembly process, at end-of-line stations or on development test benches.
Through continuous monitoring and testing, defects that would otherwise go unnoticed are detected and documented.
Five steps to AI-based noise detection
How does Sounce work?
Use-based offer model based on a software-as-a-service solution. The modular cloud infrastructure enables flexible use in various application scenarios with the option of usage-based billing.
Where can Sounce be used?
- Functionality testing
Detection of anomalies in moving components and systems in series. Material testing
Self-learning and automated testing of materials by means of vibration measurement.Acoustic assessment
Combination of classical techniques with deep learning for reliable noise detection.
Product overview
Consultation on selecting the right solution methodology, support for series application through operations expertise – initial analysis of existing data.
At the beginning of our projects, non-binding information is provided on the possible use of Sounce on your test benches. We support you in finding connections between pieces of your existing data in order to draw conclusions about possible damage or defects.
Do you already have in-house AI expertise and are able to develop, train and adapt models to your specific use cases?
Sounce offers you the opportunity to turn your AI models into a functional software-as-a-service solution that can be quickly and easily integrated into your processes.
In addition, you can benefit from our AI application expertise and increase the accuracy of your AI models in the application on the test bench.
Do you have existing test processes with defined limits? Sounce offers you the ability to complement these fixed limits with automatically self-learning limits.
From model development and model training with your data to the application itself and the associated monitoring, Sounce covers all the functions needed for production use.
Use cases
Functionality testing
Quality assurance in motion: Deep learning for functional tests of components and component groups.
Quality assurance software must be reliable while also being able to adapt flexibly and quickly to new demands.
Sounce enables intelligent monitoring of mechanical component tests, for example, when checking motion, positioning or alignment. The actively learning system also detects newly occurring faults precisely and in good time. The use of deep learning avoids the need for time-consuming work on defining manual limit values and also provides new insights into error characteristics and their cause.
Material testing
The quality and behavior of materials is determined by means of various test methods in order to ensure high-quality processing.
A large amount of test data and process parameters are often available without any direct noticeable correlation to material quality. Subsequent destructive testing is expensive and time consuming.
With Sounce, information from different data sources can be quickly and effectively collected and analyzed, meaning that faults and defects can then be detected automatically. The self-learning system evaluates test processes, which can also be monitored remotely via the web application. Possible uses include monitoring of welding, milling or injection molding processes.
Acoustic assessment
A new way of approaching acoustic testing: A combination of conventional techniques and deep learning to detect noise both today and tomorrow.
Testing and compliance with acoustic quality specifications is becoming increasingly important. Defining assessment systems for individual noise types or indicators is complex and time consuming.
Sounce offers acoustic testing without the need to pre-define or specify the anomalies to be detected. For example, in machine-speed-dependent assessments, combining conventional analysis techniques with deep learning methods offers new ways of reliably detecting noise and identifying its cause.
Our partners
Strategic partners we work with in the Sounce sector
Gain more insights on how Sounce by MHP leverages the capabilities of Amazon Web Services (AWS), why the transformation of quality testing processes contains enormous potential, and how Artificial Intelligence can be applied to sound patterns.
