Formant Synthesis

Notes
audio

Graph

Steps

  • Input - Phoneme
  • Specify formants
  • Generate source signal
  • Apply filters (formants)
  • Synthesized Speech

Strengths

  • Fully Controllable
  • Tiny footprint
  • can generate any sound

Limitations

  • Robotic
  • Require expert knowledge
  • Hard to model coarticulation
  • No natural prosody

Example

  • DecTalk used by Stephen Hawking

References