next up previous contents
Next: Handwriting Synthesis/Speech Synthesis: Up: Handwriting-speech control Previous: Handwriting Recognition/Speech Synthesis:

Speech Recognition/Handwriting Synthesis: The user dictates a ``written'' letter

Speech musculature, vocal chords, respiratory system: vocal sound production
Microphone, speech recognition algorithm
Synthesized handwriting in the form of pixel ink, or a handwriting trajectory production animation is presented to the user
The user reads the handwriting from the screen, directly at the time of spoken text dictation, and at a time of later reading, potentially also by another user

In this form of bimodality, spoken words are transformed into synthesized written text. It is unclear whether the underlying algorithm may make use of the fact that handwriting is in fact richer, than an intermediate representation like ASCII is. Handwriting does contain some parameters, like pen pressure, speed, or character size and slant, which could be inferred from the spoken text. Examples: loud in speech big in handwriting, hasty speech handwriting more slanted to the right. A writer-dependent handwriting synthesis model costs much time to be developed, although better tools and models are more and more available [303,271]. Depending on the type of handwriting synthesis model, shape simulation or combined shape and movement simulation can be performed. The first type (shape only) is already available in commercial form.

Esprit Project 8579/MIAMI (Schomaker et al., '95)
Thu May 18 16:00:17 MET DST 1995