If you have any suggestion or feedback when using MS TTS on Vista or Windows 7 or TTS APIs, this blog is a place you can post your comments about:
1. issues you find when using TTS in current products
2. suggestions you have in mind that you want in next version of the product
I promise I will read and response to each comments. :-))
There are 3 things in Vista:
1. When using Narrator the Microsoft Anna voice is too too slow. Try reading a document with the cursor keys. The voices lags a lot. You press a key and it responds after half a second. You might say that this delay is nothing. But half a second here and half a second just a second later is too much. Especially if you are blind and you are trying to use the product to work on your computer.
2. Natural voices are generally too slow. They are slow to respond, slow to "think" what to respond (what their atterance will be) and too slow in their speech even at high speech rates. Even if you change their speech rate, they generally put too many pauses between word, I guess trying to sound like humans who think in between their phrases. But if you are blind and you are trying to use such a voice like Microsoft Anna for a whole day, then this so called naturalness it drives you grazy.
3. Microsoft Anna is much better sounding and intelligible than Microsoft Sam but it is still not state of the art. Try the Nuance Realspeak voices for example, like Daniel, and compare. In so many year I expect you to do more. Factor in the horrible delays that Narrators puts in and it makes a system very hard to use.
4. Narrator lack important features found in other OSes basic screen readers, like HTML texts navigation. It is annoying, announcing by default for example all the key one presses, e.g. it says tap every time you tap somewhere, which might be nice to hear when pressing letters for typing in an edit box but so bad to hear when using say tap to navigate a dialogbox. It also has horrible keyboard shortcuts, like insert+ctrl+g that do stupid and unecessary things like "read next sibling of current element". What? Who will understand and use such a poorly named and unnecessary feature? The shortcuts themselves are also poorly chosen and hard to remember.
5. What about other languages? Where are the text-to-speak voices in Windows supporting other languages? And if you don't have the money or time to make them, which you should have, then why don't you provide a kit for developers to easily extend the current natural voice in Vista to add more language support?
If this technology were to implement in a manner that's economical, reliable, imagine how many people it could help.
Thank you sharing.
Thanks you very much that you raised many detailed feedback. Some are related to Narrator, some are related TTS engine.
I will forward your comments about narrator to narrator team to help them thinking on the design.
For your comments on TTS, yes, naturalness sounding, pause and break, language coverage etc are what we need to tackle with.
For the kit for developers, what kind of kit do you have in mind? Developing TTS for a new language requires many effort on natural language processing and wave recording. It is not so easy for individual developers yet
thanks for sharing your feedback.
The technology is easy to be used on Windows Vista and above with SAPI or managed speech API which you can refer to my previous post.
what is the "reliable" you mean?
How do you add a new TTS voice and select it to be the default ?
This was easy in XP and there is a work around for Vista. No one seems to know how to do this in Windows 7.
I do know that the Voice information must be entered in the HKEY_LOCAL_MACHINE ... Speech Voices Token section of the registry, but some of the fields are undefined.
It would be helpful to know specifically what is needed for each field or if there is some programmatic means of setting the default Voice.
We have multiple Voices installed on a Windows 7 64 bit system. The following routine shows all of the Voices:
Dim Token As SpeechLib.ISpeechObjectToken
For each Token in Voice.GetVoice
The list box in the Control Panel shows only a single entry: Anna.
Thank you very much.
Do you install 32 bit TTS on 64 bit Win7?
If so, you need to run the .net app as 32 bit process
Thanks for this page.
How does one use 32 bit TTS voices on 64 bit windows 7?
I can see the list of voices when I run the 32 bit tts properties panel from Windows\sysWOW64\speech\SpeechUX\SAPI.cpl. However, I am unable to preview any of the voices.
How can preview and set one as default?
The 32 bit voice should be able to be used in 32 bit program. Have u tried that?
if you run 32 bit control panel from windows explorer.
the 32 bit voice should be able to show and speak as well
could you send me a french vocal for my narator window 7
There is no french TTS on windows 7.
Is there a chinese TTS on Window 7. If yes, where can I download the voice
I think if you install Win7 with Chinese language pack, you can find Chinese voice.
To say it by short means. You people have A LOT of money to work with, and still your text to speech engines pretty much sucks. I'm not being an *** or anything but how hard would it be to buy some TTS company so Windows users could benefit from the technologies of for example NeoSpeech (just my own favorite, I know there are many good TTS engines out there)
Microsoft Sam... He was horrible, Microsoft Anna is acceptable, but Neospeech Julie (or Paul, dosn't matter) is really good !
Try it out:
Thanks for the feedbacks
Sam, Anna are based on old technology. You may try some latest server voice here
Let us know your feedback about the quality.