A hybrid time-frequency domain approach to audio time-scale modification

Frequency-domain approaches to audio time-scale modification introduce a reverberant or phasy artifact into the time-scaled output. Such artifacts are generally not present within time-domain implementations. However, high-quality time scaling in the time domain is typically limited to quasi-periodic signals such as speech. A hybrid method of time scaling is presented, which draws upon appealing aspects of both time-domain and frequency-domain implementations. The technique described can be applied successfully to a wide range of audio and is both robust and efficient. Subjective testing demonstrates that the technique reduces the presence of phasiness significantly.