Low Latency Audio Pitch Shifting in the Frequency Domain

This page illustrates a low latency audio pitch shifting algorithm working in the frequency domain, the "Ocean" pitch shifting algorithm. This algorithm, working with the STFT (but unrelated to the phase vocoder) will be presented and published at/by the IEEE International Conference on Audio, Language and Image Processing (ICALIP). The main particularity of the algorithm is to be very resistant to reduction of the FFT size. Hence it allows very low latencies to be achieved with only minor degradations in term of quality.

The following tables illustrate two tests (pitch shifting by +10%) on selected audio excerpts.

Test 1 (Highest quality settings)
Original version Improved phase vocoder Ocean algorithm SOLA
Listen Listen Listen Listen
Listen Listen Listen Listen
Listen Listen Listen Listen
Listen Listen Listen Listen

Test 2 (With latency constrained to be below 12ms)
Original version Improved phase vocoder Ocean algorithm SOLA
Listen Listen Listen Listen
Listen Listen Listen Listen
Listen Listen Listen Listen
Listen Listen Listen Listen

Sources

The excerpts are taken from the following musics:
Music 1: The Chariots of Fire, by Vangelis, in Odyssey: The Definitive Collection
Music 2: La Danse Macabre, by Saint Saens, in Synthétiseurs, les plus grand thèmes classiques, vol. 5, adaptation: Ed. Starink
Music 3: Cleanin' Out My Closet, by Eminem, in The Eminem Show
Music 4: The Caves of Skops (Track 41), by Ubisoft, in Rayman 1