Welcome! The talking moose was famous 25 years ago.
A lot has changed since then, with computing power and the internet. I think the time has finally come, to use advances in 3D graphics and speech synthesis, to make a fantastic new Talking Moose.
I’ve been slowly developing this new Moose over the past few years, aided by modern programmers, and now I’ve decided to share with you, the story of developing the new Moose. The development process, along with my mistakes and setbacks, you may find interesting.
I think the new Moose will be incredible. Not as revolutionary as the original moose was in 1986, but I truly believe that this new 3D character, with his new focus on humor, usefulness and internet content, will show the world a successful formula for a tolerable talking character/agent.
On the right-hand sidebar, are links to blogging topics. Click those topics to start reading about this new Moose.
Below, are some of the latest posts, but kindof random. A better way to learn more, is clicking on the Blogging Topics to the right.
Several years have past since I last had a coder working on a shadow behind the moose. It will soon be time to re-start the coding of a shadow. This post will simply be used to store some notes for that upcoming coding task. Continue reading “Shadow notes” »
I just bought from Cereproc.com, the cerevoice Hannah. It works great, (and so does the William voice).
If you’ve been following this development blog, maybe you noticed that my youtube videos just improved in quality. I changed Bandicam to use these parameters: 640×360, H.264 encoder, 44.1 audio at 192kbps, framerate 59.94fps.
This post documents the difficulties we had, due to poor Microsoft documentation of the C# workings, of Phoneme IDs, that might be simple Phone IDs numbered 1 to 49, or unicode value of IPA (International Phonetic Alphabet), or UPS format in unicode. The Talking Moose’s lip sync animation uses visemes, not phonemes, but we encountered quite a few Voices that had minor errors in their visemes, and so, to correct those errors, we decided to look at the associated phoneme, to help choose a more correct viseme to display. What a can of worms. Continue reading “PhonemeReachedEventArgs – Getting Phoneme events from SAPI speech synthesizer to work” »
I couldn’t find this little error documented anywhere easily findable with a Google search, so I’m making this post to help others in the future, who work with SAPI voices doing lip-sync animation with Viseme IDs.
But first, here’s the Talking Moose speaking a joke using Microsoft Anna’s voice.
The Microsoft Anna voice is the default voice on Windows 7, US English. Viseme ID = 21 is supposed to be for “m”, “b” and “p” consonants. Unfortunately, Anna voice sends a viseme ID = 19 for the letter “b”, which is wrong. I’m going to have to code an extra test for this voice and this viseme, to fix it. bummer. (If I don’t fix it, when the Moose speaks “Bob”, it looks like “tot”,”dod”,”non”. Since “b” is a highly recognizable lip shape, I want to make sure users see it.