home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.amiga.audio
- Path: sparky!uunet!sun-barr!ames!ncar!unmvax!mimbres.cs.unm.edu!nmt.edu!aevans
- From: aevans@nmt.edu (Alex Evans)
- Subject: Re: Halving sample speed while keeping pitch
- Message-ID: <1992Aug27.020717.26929@nmt.edu>
- Summary: Two methods
- Keywords: DSP FFT Absolute Frequency Shift
- Sender: Theodore (Alex) Evans
- Organization: New Mexico Tech
- References: <9208241809.AA00474@.nairobi.inel.gov.inel.gov.> <1992Aug24.194501.21918@dgbt.doc.ca>
- Date: Thu, 27 Aug 1992 02:07:17 GMT
- Lines: 23
-
- I can think of two possibilities to satisfy the requirements
- of the original poster, to slow down speech without changing the
- frequency.
- Since a single persons speech in English does not typically vary much
- in pitch it is possible to get fairly decent results by multiplying the slowed
- down sample by a sinusoidal wave that has a frequency equal to the difference
- between the original and the new pitch of the voice. For example if someones
- speech averages at 2000Hz and you slow it down to 75% of its original speed
- you would multiply the signal by a 500Hz sine wave. This would cause the
- voice to be heard at both 2000Hz and 1000Hz, if the lower frequency is a
- problem you can filter it out.
- The other solution that comes to mind is more complicated, but retains
- harmonics. Run the signal through a FFT. Stretch the resulting array by the
- amount you want to slow it down. You now have two choices. Either run this
- array through an inverse FFT, and play it back at the original sample rate.
- Or only take the lower points producing an array the size of the original run
- this through an inverse FFT, and play back at a reduced sample rate.
-
-
-
- --
- Theodore (Alex) Evans aka Alexei Vnebrachnei Ivanovitch
- MOL College of St. Golias
-