TTS and the Lack of Good Documentation
Part of my job is to evaluate applications that may add value and improve cost effectiveness within the arena of eLearning. Recently we explored using Text to Speech (TTS) as a means of quickly and easily adding narration without finding voice talent.
Mileage really varies; it seems TTS is meant either for screen reading, or telephony systems. It doesn't seem to be well marketed as a viable option for eLearning development. There;'s no standard for comparing them, and few ties that even try to compare a few, so it's a big time sink to research.
TTS comes in two flavors; robotic and "natural voice" although most natural voice is still pretty choppy and synthetic. When deciding on purchases, know that when you choose a natural voice option, you'll most likely need to purchase (and program) voices separately.
The two best options unfortunately are pricey. NeoSpeech is very popular, but is still choppy, and the quality degrades quickly when played on slower systems. Loquendo is eerily close to human speech, and is built to add in voice effects that humanize the text. They even have voice effects that can be used, including types of laughs, coughs, and more. If you're curious, go to their site and play with it.
Unfortunately, Loquendo is pricey. And extremely slow in responding to requests for pricing. Once Loquendo finally got around to responding to my requests for pricing and trail software, the price was exorbitant (12,000US), and still involves annual fees, even though we aren't resellers. It wasn't worth getting approval to use the trail version, which you cannot download without discussing use and price with them first.
NeoSpeech's VoiceText is more affordable, but at 1,500US annually, they don't seem to care much about the users. It's user interface is barely more than a shell, and you have to open the user dictionary separately to create custom pronunciation. The documentation is some of the worst I've seen, as it doesn't include installation and feature descriptions, instead opting to focus on coding, which is great.. if you're already a TTS and telephony expert. All that for something that still sounds choppy? No, thanks. Someone really needs to revamp that documentation.l
Overall, it's been like looking at a vista through a pinhole. I'm sure I'm not the only one who's considered TTS only to be stymied by the death of information. It seems like they don't want to explore the eLearning market.
There are several others that are more intended as reader tools for the visually impaired. Mileage really varies, both in quality and cost. Some I suspect have or are spyware. Trying to uninstall some I tested was a real chore.
But I must admit, during my more frustrated moments, I had a blast playing with them by using some lyrics from "Birdhouse in Your Soul", especially when slowing down the speech rates.
Mileage really varies; it seems TTS is meant either for screen reading, or telephony systems. It doesn't seem to be well marketed as a viable option for eLearning development. There;'s no standard for comparing them, and few ties that even try to compare a few, so it's a big time sink to research.
TTS comes in two flavors; robotic and "natural voice" although most natural voice is still pretty choppy and synthetic. When deciding on purchases, know that when you choose a natural voice option, you'll most likely need to purchase (and program) voices separately.
The two best options unfortunately are pricey. NeoSpeech is very popular, but is still choppy, and the quality degrades quickly when played on slower systems. Loquendo is eerily close to human speech, and is built to add in voice effects that humanize the text. They even have voice effects that can be used, including types of laughs, coughs, and more. If you're curious, go to their site and play with it. Unfortunately, Loquendo is pricey. And extremely slow in responding to requests for pricing. Once Loquendo finally got around to responding to my requests for pricing and trail software, the price was exorbitant (12,000US), and still involves annual fees, even though we aren't resellers. It wasn't worth getting approval to use the trail version, which you cannot download without discussing use and price with them first.
NeoSpeech's VoiceText is more affordable, but at 1,500US annually, they don't seem to care much about the users. It's user interface is barely more than a shell, and you have to open the user dictionary separately to create custom pronunciation. The documentation is some of the worst I've seen, as it doesn't include installation and feature descriptions, instead opting to focus on coding, which is great.. if you're already a TTS and telephony expert. All that for something that still sounds choppy? No, thanks. Someone really needs to revamp that documentation.lOverall, it's been like looking at a vista through a pinhole. I'm sure I'm not the only one who's considered TTS only to be stymied by the death of information. It seems like they don't want to explore the eLearning market.
There are several others that are more intended as reader tools for the visually impaired. Mileage really varies, both in quality and cost. Some I suspect have or are spyware. Trying to uninstall some I tested was a real chore.
But I must admit, during my more frustrated moments, I had a blast playing with them by using some lyrics from "Birdhouse in Your Soul", especially when slowing down the speech rates.
Comments