For about a month now I'm jobless. Which is a strange situation. I've never, ever been without a job before. After I graduated I've always been so lucky to find something, and almost always something I like. There are a few opportunities though, so I'm not at all desperate.
The fun side is that I have more time to spend on pVoice. Which is A Good Thing. pVoice 2.0 is out since mid-March, and I released it on the website. After a few days I discovered a few annoying bugs already.
It turned out that I couldn't compile in the Image::Magick DLL's, and couldn't get a statically linked version together. Therefore I had to go back to using wxPerl's own Image handling routines. That's not so bad, but I desperately wanted .WMF support. The symbol library my daughter uses is distributed in .WMF format. The only Perl module I could find that supports .WMF is Image::Magick. But I converted all 3500 images to .JPG and that can be read using standard wxPerl.
After that I discovered that pVoice 2.0 couldn't interface with the Adremo electric wheelchair my daughter uses. That was a more important bug (at least for my own situation). It turned out that I must have been sleeping while I implemented that piece of interface. All kinds of errors within that code. But OK, it's fixed now, and the current version is 2.0b.
In the mean time I've contacted a few companies that develop high quality Text To Speech engines.
The reason is that the free TTS engines from Microsoft are of medium to poor quality. I've seen it now with Krista: you want to understand what she's saying without asking 'What did you say?' three times and finally looking on her dispay to read it yourself. That's not what a communications package is about.
Commercial Text To Speech (TTS) technology is expensive. Most of them have a (for them) very lucrative strategy: first the developer has to pay somewhere between 1000 and 3000 euros just to get an SDK, with which the developer can modify his software to support that TTS engine. Then you'll have to pay royalties for each sold product that includes their technology (which varies from 10% of the end-user price to 600 euros per copy). And finally you'll have to cough up a pre-payment of those royalties yourself (somewhere between 6000 and 8000 euros). Needless to say that for free software like pVoice those amounts are not an option.
So I contacted a few manufacturers of those TTS systems to ask them if I could get the SDK for free, forget about the prepayment and just pay for the royalties. That would mean there have to be two versions of pVoice: a free one and a commercial one. The commercial one would support a commercial TTS engine, the free version keeps using the Microsoft Agent.
All TTS developers said no, except one. One said yes! I'm not mentioning the name here (although I've mentioned the name on #perl already), because the contract has not yet been signed, but that seems to be a matter of days. I won't have to pay for the SDK, I won't have to pre-pay for the licenses, I can develop pVoice 'Commercial' for all languages they support (currently Swedish, English, German and French, more to come...Dutch next month), and I got a large discount on the normal royalty-fee.
So now I'll have to wait a few days to finish the details on this 'sort-of-sponsoring' contract....I can't wait!
