Friday, October 31, 2003

Repeat After Me This hasn't been covered much, to my knowledge. In the Developer Tools install for Panther, there is a new application called Repeat After Me (in the /Developer/Applications/Utilities/ folder). Kim S often told me about this program, and that they couldn't find the source to the original version. It looks like they've overcome that issue and created an OS X version. Basically, Repeat After Me allows you to enter a string of text, generate a phoneme list, then speak the string and apply your cadence and pitch to the phoneme list... in other words, you can make the TTS engine pronounce a sentence exactly as you want it to, simply by mocking your own reading of the sentence. Here's a sound sample of the default pronunciation: And here's the result of using my voice to control the pronunciation: We wanted this feature for PuppetTime SO BAD! What we were missing was the ability to extract pitch and timing information from the audio... I wonder if these technologies will be exposed via new APIs? 11:59:13 AM    trackback []