[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Pitch Shifting Algorithm

Ben, your question is somewhat unanswerable. Since spectral and temporal structure of sounds are interlocked, and you'd like to change one without messing up the other, you need to decide exactly what aspects you don't want to mess up. Essentially, you fit a model of the sound, based on perception, modify the parameters, and re-synthesize. For ba/da discrimination, you need to fit a model that includes fast formant transitions, I suppose, with an analysis tuned to human perception and a re-synthesis tuned to rat perception.

Sorry I don't have a more constructive answer.


At 7:33 PM -0500 6/25/06, Ben Hornsby wrote:
Hi All,
I've had a request from someone in our department for suggestions for a pitch shifting algorithm that will be used to shift the frequencies of phoneme pairs (e.g. ba-da) into the center range of rat hearing (~8kHz) without affecting the temporal cues of the speech stimuli. I'm unfamiliar with available software to do this (and pros and cons for specific kinds). We do have some commercial software here such as Adobe Audition and SoundForge that have pitch shifting algorithms as part of their software package. We are hoping to limit artifact as much as possible. Would these be appropriate algorithms to use? If not, why not and any suggestions as to which software program would be more appropriate?


Ben Hornsby