I'm struggling with the Pronunciation Dictionary...
1) Microsoft Zira and David will speak phonemes in the "New Dictionary Entry" box. Hazel will not. Perhaps a problem between SAPI4 and SAPI5, but I cannot find which API applies to which voice...
2) The Phoneme Helper gives "S1" as primary stress marker, but testing the phoneme returns an error: Speech error: SSML input not valid; cannot find "speak" element.
3) My PC has additional Microsoft voices (Susan, George, Mark) that do not appear in TA. How do I add/activate them?
4) I used to use Acapela, but don't anymore. How do I delete a voice?
Sorry for the late reply, I'll try to help with your questions ...
1. The Microsoft voices you mentioned (that appear in TextAloud) are Sapi5 voices. I'm not sure why Hazel is having trouble with phonemes. It may help if you send a TextAloud problem report. From the TextAloud menu click Help -> Support -> Report A Problem, and mention this issue in the problem report text. System information included with the problem report may give me some clues as to what is happening.
2. In the problem report above, please include which voice you're seeing the speech error with when testing the stress marker. I'm not able to duplicate myself yet. It may also help if you copy and paste the exact phoneme string into the problem report text.
3. Are you running Windows 10? If so, the Susan, George and Mark voices may show up in your Windows 10 text to speech settings, but not in TextAloud. The reason is that voices displayed in Windows 10 settings use a different speech api. They are not Sapi5 voices, and TextAloud does not currently support them.
4. To delete your Acapela voice, you should be able to uninstall the voice using the same Windows panel in Windows Settings that you use to uninstall other applications.
Jim Bretti wrote: ↑Fri Mar 09, 2018 1:24 pm
Sorry for the late reply, I'll try to help with your questions ...
1. The Microsoft voices you mentioned (that appear in TextAloud) are Sapi5 voices. I'm not sure why Hazel is having trouble with phonemes. It may help if you send a TextAloud problem report. From the TextAloud menu click Help -> Support -> Report A Problem, and mention this issue in the problem report text. System information included with the problem report may give me some clues as to what is happening.
2. In the problem report above, please include which voice you're seeing the speech error with when testing the stress marker. I'm not able to duplicate myself yet. It may also help if you copy and paste the exact phoneme string into the problem report text.
3. Are you running Windows 10? If so, the Susan, George and Mark voices may show up in your Windows 10 text to speech settings, but not in TextAloud. The reason is that voices displayed in Windows 10 settings use a different speech api. They are not Sapi5 voices, and TextAloud does not currently support them.
4. To delete your Acapela voice, you should be able to uninstall the voice using the same Windows panel in Windows Settings that you use to uninstall other applications.
I first deleted the Acapela voice... and things changed elsewhere, for the better. Microsoft Hazel now works fine, including phonemes. Dave and Zira still return an error, but since I prefer Hazel, I don't much care. I'll send a problem report anyway.
I don't hear much stress when using the stress marker... but at least there's no error (with Hazel). The word I was hassling with is "facade", phoneme [F AX . S1 S AA D S1] with the second syllable stressed. This phoneme is pretty close.
I am indeed running Windows 10, so thanks for your explanation about Susan/Mark/George.
Thanks for your help. I'm looking forward to V4. I first started using TA maybe twenty years ago, so I guess I can afford to buy another copy...
Thanks for sending the problem report. It looks like you sent the report after removing the Acapela voice, but the fact that removing Acapela changed things actually could make sense if you had an older version of Acapela installed. If you send an email to support@nextup.com they'll send you current links for Acapela voices you purchased. Let me know if you try that and run into the same issues.
The phoneme problem you're having is that F (upper case) is not a valid phoneme for US English. US English and UK English use different phoneme sets, and phonemes are case sensitive. So for US English, you could try starting with something like this: [f ax s aa d]. Note that lower case f should work. The phoneme helper dialog available in dictionary maintenance provides a list of phonemes available for the language assigned to your dictionary. The Test button beneath the test text box is not validating the phoneme string, which explains the error you're seeing when you use that test button. In the next update I'll make sure both test buttons are validating the phoneme string.
For the stress marker, you might try placing the stress marker in different positions in a syllable to see if makes any difference.
