Your posts made me reread my paper, and indeed the DL were smaller than 0.5%. They were for complex tones 0.36% for F0=100Hz, and 0.19% for F0=250Hz (cut frequency of the high pass 1 kHz, no low-pass masker). This is in accordance with what has been reported in the literature. Pure tone DL were much higher (0.43% for 250 Hz, 0.89% for 100 Hz). I suppose that F0 frequencies below 1kHz are quite relevant to music so that relatively high DLs (>0.1% for complex tones) should be taken into account when dealing with music.
Best, Chris
Date: Wed, 14 Jan 2009 16:14:37 -0600 From: "James W. Beauchamp" <jwbeauch@xxxxxxxxxxxxxxxxxxxxxx> Subject: Re: f0 frequency difference limen In terms of musical intervals, 0.2% for partials, 0.1% for the F0 translates into 3.4 and 1.7 cents, respectively, where 100 cents corresponds to an equal-tempered semitone change of pitch. I wouldn't think that this result would be completely independent of F0. I would expect the percentage threshold to increase for F0s below a certain point. It would also depend on the spectrum. I.e., spectra with just a few partials in the bass region may have much larger DLs. But 2 cents is a good practical df to shoot for in any music synthesis system. Jim
-- Prof. Dr. Christian Kaernbach Allgemeine Psychologie Institut für Psychologie Christian-Albrechts-Universität zu Kiel Olshausenstr. 62 D-24098 Kiel Germany www.kaernbach.de