Currently, the equalizer settings for media playback via a headset are selected based on a limited number of user studies. Equalizer preferences depend on various factors, including earbud fit (which impacts audio leakage levels) and the genre of audio being listened to. This disclosure describes techniques to automatically set or recommend equalizer settings using machine learning models that are trained on a crowdsourced dataset of user preferences, obtained with user permission. Equalizer settings, audio genre, and magnitude of acoustic leakage are obtained from deployed media playback devices and are used to train the machine learning model. During operation, the trained machine learning model adjusts the equalizer to a setting optimized over observed consumer preferences.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.