• About
  • Tour Plan
  • Discography
  • Blog
  • Peter’s Lab
  • Video Clips
  • Playlist Compendium

Peter Williams

Musician – Bassist – Composer

Feeds:
Posts
Comments

Speech Feature Extraction

February 10, 2016 by Peter2708

https://peterwilliams.dk/wp-content/uploads/2016/02/WIN_20160215_21_56_03_Pro.mp4

For this project I designed an audio based computer game. The player controls the environment with their voice. The Pure Data patch that I wrote is able to distinguish between various vocal sounds.

  • SSSSSSSS
  • ZZZZZZZZ
  • Open vowels
  • Hard constanants

These four sounds were used to drive an external physical model of a D.C. motor.

Saying ‘ssss’ or ‘zzzz’ is effectively putting your foot on the gas, say ‘ka’ or ‘pa’ changes up a gear. ‘Aaaah’ turns on reverb and a ping ping delay to simulate other traffic, and, for the icing on the cake, ‘sssss’ switches on a randomly sweeping notch filter, the center frequency of which jumps further, with increased time intervals, the louder the input voice. This has the effect of making the motor sound like a speeding boat. ‘Zzzzz’ switches off the notch filter, and engages a three tap dynamic chorus effect, again controlled by amplitude. This makes the motor sound like a plane.

Despite making people look quite silly, it proves to be quite an addictive waste of time.

Screen shot 2016-VocalExtractionScreenShot02-15 at 21.42.45

Posted in Peter Williams, Peter's Lab | Leave a Comment

  • all the important links

Get a free blog at WordPress.com

Theme: Mistylook by Sadish.