- The simulations are about three sounds i (as in keep) , u (as in loot), and a (as in father).
- 10 versions of each were recorded on a PC with a headphone with speakers (really lowend), from one person over 5 days producing 30 .wav files.
- Those files were read by matlab, and translated from the time to the frequency domain by fourier transform.
- This was then translated to mel-frequency cepestral coefficients.
- The data was also normalised for volume, so that all of the sounds had the same volume.
- Neurons were stimulated for 3 cycles.
- keep1.wav
keep2.wav
father1.wav - Roughly speaking, this is how hair cells fire, so this is biologically realistic input.
