Quality of Microsoft TTS 5.1September 25, 2008
I'm currently downloading Microsoft's TTS (Text To Speech) API version 5.3, which was release within the last year.
I sure hope the quality of the new engine is better than the one in version 5.1
Have a listen to the synthesis of "What is tomorrow's weather forcast"? Yuck! Granted, this is using 2001 technology.
Exercise 19September 25, 2008
SummaryCreate an application which can, upon hearing a specific voice command, reply by speaking a specific phrase.
SolutionMicrosoft's Speech API makes it surprisingly easy to add both voice recognition and text to speech to an application. I was hoping to use version 5.3 of the API, but after downloading and install Windows SDK 6.1, which includes Speech API 5.3, it wasn't listed in Visual Studio 2002 in the "Add References" dialog.
I ended up using version 5 of the API, which is all the way back from 2001. The text-to-speech quality is brutal, but the programming steps would be about the same regardless of which API you use.
DownloadClick hereSpeech recognition. etc.September 24, 2008
Ever since Meredith got home from Seattle and told me about "Grace", the AI that runs the Microsoft Home of the Future, (I love the name Grace as the name of an in-home AI agent!) I seem to be drawn in by the promise of creating a full spectrum AI system with the following components:
  | A speech recognition layer |
| A language parsing layer |
| A core knowledge representation |
| A deduction engine |
| A web interface for looking up well defined information such as the weather via web services |
| A voice synthesis layer |
And to top it all off:
A subtle, well designed microphone, speaker, and LCD installation in the kitchen.
I'm not really interested at this point in working on the voice recognition or voice synthesis layers, I just want a cheap, but easy to use and ultimately effective system.
Here are some fun use cases to think of:
| "Grace, what is the weather forecast for tomorrow"? |
|  | theweathernetwork.com would load on the LCD for Waterloo, ON. |
| | Grace would speak the weather forecast. |
| "Grace, how many new emails do I have?" |
| | My GMail account would load on the LCD. |
| | Grace would speak the number of unread emails. |
| "Grace, what time is it?" |
| | Grace would speak the current time. |
| "Grace, what is my Social Insurance Number?" |
| "Grace, what is aunt Beryl's phone number?" |
| "Grace, what is the police phone number?" |
A little bit more out there:
| "Grace, order meal #1 from swisschalet.com" |
| | Grace would speak the details of meal #1 and ask for confirmation. |
| | Grace would speak an update once the meal was ordered, and would indicate the expected delivery time. |
etc. How fun would that be to tinker with!
older >>
