Ensuring clear speech intelligibility is one of the biggest challenges for professional sound engineers in television, streaming, and live productions. Algorithms developed by Fraunhofer IDMT in Oldenburg, Germany, are analyzing listening efforts in real time and can provide adjustment recommendations.
Furthermore, the company’s digital assistant, “YourSound,” allows users of audio devices and infotainment systems to playfully customize the sound to their personal preferences.
And, the researchers are not only focusing on software: Fraunhofer IDMT is bringing AES67 to DSP hardware for audio over IP.
Unclear dialogue in a television series, excessively loud background music in reality TV shows, or a sports interview drowned out by audience cheers can all detract from the viewing experience. Technical measuring devices are used in media production to monitor levels and volume. However, an objective method for measuring speech intelligibility still needs to be established.
Production professionals in radio, television, and streaming should be able to ensure that dialogues are always clearly understandable to the audience. Therefore, the Fraunhofer IDMT-HSA in Oldenburg developed the “Listening Effort Meter” (LE-Meter) for the objective evaluation of speech intelligibility. The technology is available under license and is used by various market participants.
Speech intelligibility measurement has already been integrated into post-production software and implemented as a plug-in. To this end, Fraunhofer IDMT-HSA partnered with a leading streaming platform to bring the optimal post-production solution to market. Technically, the measurement is based on AI algorithms and has been extensively evaluated in recent years. Beyond measuring speech intelligibility, further functionalities are conceivable, such as automatically marking critical sections directly in the timeline or automatically improving speech intelligibility.
Researchers at Fraunhofer IDMT-HSA have found a simple and elegant way to personalize sound and dynamics in a wide range of applications. The “YourSound” technology has already been implemented in headphone software and a multimedia system for vehicles. The developed audio software allows for individual sound settings without requiring users to navigate complex submenus or parameters. Users of audio devices, such as multimedia platforms or smartphones, can easily adapt the sound to their personal needs and preferences.
The technology uses a presentation of musical examples that users can customize to their preferences via a simple user interface. Once set, the individual presets positively influence the overall sound. This can lead to a better listening experience, regardless of the volume. Thanks to the new algorithms from Fraunhofer IDMT-HSA, the sound of music and films is thus adapted to individual listening preferences.
With the “Minikraken” headphone amplifier demo from Fraunhofer IDMT, multi-channel audio can be routed, distributed, and recorded via standard network switches instead of conventional audio cables. Fraunhofer IDMT-HSA has developed a fully functional software stack based on the AES67 standard, built on cost-effective DSP hardware, which can be used in various audio applications. AES67 is an open interoperability standard from the Audio Engineering Society (AES) for professional Audio over IP (AoIP), designed to connect various proprietary audio network systems (such as Dante, RAVENNA, and Livewire). The DSP can process at least 16 input and output channels. Additional signal processing on the DSP can be added. Depending on the configuration and network architecture, latency ranges from 0.75 to 5 ms. A minimalist web application and discovery implementation allow users to find other devices on the network, connect to them, and configure settings.
