Sixth International Conference on Spoken Language Processing
This paper presents a technique to remove the hum in speech corpora based on time domain subtraction rather than spectral subtraction. This novel method is able to avoid any perceptible or measurable artifacts and has proved to be an efficient technique for completely eliminating even complex hum waveforms from speech recordings.
Bibliographic reference. Pfitzinger, Hartmut R. (2000): "Removing hum from spoken language resources", In ICSLP-2000, vol.3, 618-621.