Show Posts
1
Speech Technology / Three problems with Ultra-Hal TTS
« on: June 04, 2005, 10:26:21 am »
Being outside the US for a period, I have been using Ultra-Hal to read transcripts from the Rush Limbaugh program's website while I'm doing work around the house. However, I regularly run into three irritating problems:
1) When the program is set to read the text in windows and messages that come up, if I click on something before allowing it to read the entire current message, it alters my Windows volume settings, turning WAV almost all the way down, and the balance all the way to the left. I must then manually restore the proper settings.
2) I am frequently unable to record to a .WAV file all the text I wish to at once. If I try to do more than about ten Office pages of text, Hal will record about thirty-three minutes of it and quit. I would prefer not to have to split up text and record it in two parts.
3) Sometimes Hal "records" an entire selection, and when I go to play it, there is only silence. I have no way of knowing whether it's really recording until I try to play it afterward (or I load the .WAV file into Audacity for conversion and see a flat green line indicating the file's non-existent sound level throughout.
I think the cost of ordering the lifelike voice of "Paul" is reasonable, and would be helpful for reading political articles and news show transcripts to me while I'm commuting after my return to the US. I intend to order the new voice if I can get these problems solved. Otherwise, the hassle of splitting text for recording in two parts, or rerecording a selection until it "takes," makes it all more bothersome than worthwhile.
I'm using Windows XP on a Compaq 2135US laptop.
1) When the program is set to read the text in windows and messages that come up, if I click on something before allowing it to read the entire current message, it alters my Windows volume settings, turning WAV almost all the way down, and the balance all the way to the left. I must then manually restore the proper settings.
2) I am frequently unable to record to a .WAV file all the text I wish to at once. If I try to do more than about ten Office pages of text, Hal will record about thirty-three minutes of it and quit. I would prefer not to have to split up text and record it in two parts.
3) Sometimes Hal "records" an entire selection, and when I go to play it, there is only silence. I have no way of knowing whether it's really recording until I try to play it afterward (or I load the .WAV file into Audacity for conversion and see a flat green line indicating the file's non-existent sound level throughout.
I think the cost of ordering the lifelike voice of "Paul" is reasonable, and would be helpful for reading political articles and news show transcripts to me while I'm commuting after my return to the US. I intend to order the new voice if I can get these problems solved. Otherwise, the hassle of splitting text for recording in two parts, or rerecording a selection until it "takes," makes it all more bothersome than worthwhile.
I'm using Windows XP on a Compaq 2135US laptop.