Audio tech adjusts speech to surrounding noise levels
Mon 1 Feb 2016
German scientists have devised an audio system [PDF] which promises to significantly improve the audibility and comprehension of spoken messages – for notifications in busy train stations, speakers in lectures, or even mobile phone conversations on a noisy road for example.
The Hearing, Speech and Audio Technology Project Group based at the Fraunhofer Institute for Digital Media Technology (IDMT), has developed the specialised software, ADAPT DRC, which is able to continuously analyse ambient noise levels via a microphone, and adjust speech volume in real-time.
“It is not enough to simply make the voice louder over the loudspeaker or mobile phone to drown out the noise,” explained team lead Jan Rennies-Hochmuth. He noted that “speech is much more complex,” than simply making a voice louder – this existing technology does not necessarily make the messages any clearer, with extremely loud speech projection causing ‘rattle’.
A release detailing the research today, described the importance of pitch and frequency to target the listener effectively. While vowels are generally spoken at lower pitches, and are ‘drawn-out’ and easy to understand, consonants are snappy with higher frequencies and are therefore harder to understand in a noisy environment.
To tackle these audibility challenges, the researchers designed algorithms which prioritise certain frequencies and place emphasis, at the right time, on parts of speech which would typically be problem areas for listeners in busy settings.
The software is also able to modify parts of a message which are spoken at different volumes. As speech contains both loud and softer spoken sections – termed voice dynamics – the technology can subdue louder parts while amplifying quieter bits. This process is known as Dynamic Range Compression (DRC)
The DRC system has already been developed and tested, achieving application maturity and has been made available to industrial partners. The team has argued that as phones and other loud speaker equipment already possess in-built microphone technology, the system could be easily applied to many existing setups without additional installation costs.