The aim of this work is to analyze the performance of Apple’s iPhone and iPad as voice recorders, while at the same time finding algorithms to enhance speech recordings and reduce the noise introduced by the low quality built-in microphone. We perform spectral analysis of silent recordings to acquire the noise print from different device models. Comparing these results, we assess whether or not different iPhone devices can be modeled with the same noise source. We also propose a speech enhancement algorithm to reduce additive noise introduced during the recording. Finally, we comment on the results and make a few considerations for further developments.
Full text