Speech-to-Text

kip

Member
What is Pleco's plans to offer a English Speech-to-Text and Mandarin Speech-to-Character add-on?
 

mikelove

皇帝
Staff member
kip said:
What is Pleco's plans to offer a English Speech-to-Text and Mandarin Speech-to-Character add-on?
It's already available on Android as a free, built-in feature, basically because Google supports it as an integrated Android function which took us all of 10 minutes to implement.

We expect that Apple will be doing the same thing as they expand Siri support to Mandarin - already just added it for Japanese - and once they do we'll be happy to hook into that system as well. But licensing a separate system and offering it as a paid add-on doesn't really make sense given the likelihood that Apple will soon make it possible for any app to offer speech recognition of at least as good quality for free.
 
only problem is you have to be online for the speech to text feature to work. not always possible if coverage isn't great. i was surprised actually when I sw it on Android. great feature.
 

mikelove

皇帝
Staff member
pprendeville said:
only problem is you have to be online for the speech to text feature to work. not always possible if coverage isn't great. i was surprised actually when I sw it on Android. great feature.
Thanks! Offline would be nice, but unless we can find a system that works almost as well as online does (difficult given that they've got terabytes worth of voice templates and huge server farms) it's going to be tough to get enough people to pay for it to justify the license.
 

kip

Member
Lonely Planet and Jibbigo both offer off-line attempts of Speech-to-Text. To me, they are more musing than useful. I had hoped that Pleco in my initial inquiry would permit me to mothball those apps like the other dictionary attempts I had. It turned out to be a feature too far (at this time). *** For what Pleco CAN do, I give kudos to the designer(s) and programmer(s), an excellent job peppered with pleasant surprises. Many "Nice!", "Sweet!", "Slick!", etc moments. ***

What are Pleco's plans to offer full sentence/paragraph translation? (including back translation for better translation confidence)
 

mikelove

皇帝
Staff member
kip said:
*** For what Pleco CAN do, I give kudos to the designer(s) and programmer(s), an excellent job peppered with pleasant surprises. Many "Nice!", "Sweet!", "Slick!", etc moments. ***
Thanks!

kip said:
What are Pleco's plans to offer full sentence/paragraph translation? (including back translation for better translation confidence)
None, at the moment - that really does need to be done online, but Google Translate charges third-party apps a per-character fee to run translations through their service, and since they offer the exact same service in their own dedicated Google Translate app for free, we don't really think enough people are likely to be interested in our version to pay for it.
 

numble

状元
mikelove said:
kip said:
What is Pleco's plans to offer a English Speech-to-Text and Mandarin Speech-to-Character add-on?
It's already available on Android as a free, built-in feature, basically because Google supports it as an integrated Android function which took us all of 10 minutes to implement.

We expect that Apple will be doing the same thing as they expand Siri support to Mandarin - already just added it for Japanese - and once they do we'll be happy to hook into that system as well. But licensing a separate system and offering it as a paid add-on doesn't really make sense given the likelihood that Apple will soon make it possible for any app to offer speech recognition of at least as good quality for free.
Does iOS let you hook into the system and change the language you want to do speech-to-text with? For instance, if my language settings are english, would an app be able to do speech to text in japanese?
 

mikelove

皇帝
Staff member
numble said:
Does iOS let you hook into the system and change the language you want to do speech-to-text with? For instance, if my language settings are english, would an app be able to do speech to text in japanese?
Not at the moment, but I suspect they'll be adding it in OS 6 - would have been tough to do it in OS 5 since they kept Siri officially secret until right before its release.
 
I have 2 text to speech apps I use at the moment. WebReader TTS for English and PiTL Putonghua TTS for Chinese both for iPhone 4. WebReader is really good. Sounds very natural and has male and female voices built in. Putonghua TTS is very basic and doesn't sound very natural. Each word is pronounced individually so the sentences don't flow naturally compared with WebReader. Only one voice that kinda sounds female (I wouldn't fancy a date with her though). The TTS engine is adopted from Pocketsphinx. 

I'm a bit confused as to what you plan to use for your TTS. Will you use someone else's license, develop your own or use SIRI? Hopefully the one you implement will have speech that sounds natural. I think this is a priority with TTS. Also, will you need a minimum of an iPhone 4S if it's SIRI based? I've just come up with a few points which I deem are NB for good TTS. 

1. You can commence reading at any point in your text.
2. The text being read is shown in a sub-bar underneath the main text and can be shown in character form or pinyin. 
3. Can open local files in Pleco and read them and offers support for .doc & .pdf along with .txt if possible.
4. Offers Remote Control Support.
5. Multiple voice support (male and female). 
 

mikelove

皇帝
Staff member
pprendeville said:
I'm a bit confused as to what you plan to use for your TTS. Will you use someone else's license, develop your own or use SIRI? Hopefully the one you implement will have speech that sounds natural. I think this is a priority with TTS. Also, will you need a minimum of an iPhone 4S if it's SIRI based? I've just come up with a few points which I deem are NB for good TTS. 
We licensed a very good one that generates natural-sounding voice and does so offline - should work fine on any Pleco-compatible iPhone.

pprendeville said:
1. You can commence reading at any point in your text.
That should be supported, yes.

pprendeville said:
2. The text being read is shown in a sub-bar underneath the main text and can be shown in character form or pinyin. 
We weren't planning to do that - having the cursor follow along in the main text seems more natural and fits better with our popup reader UI.

pprendeville said:
3. Can open local files in Pleco and read them and offers support for .doc & .pdf along with .txt if possible.
.txt certainly, but .doc and .pdf we don't have any good way to parse - I think you'd have to clip text from those to a separate window.

pprendeville said:
4. Offers Remote Control Support.
Remote control from where? Not quite following this one.

pprendeville said:
5. Multiple voice support (male and female). 
We're currently only planning to support female - their male voice isn't all that good and we feel like we already cover that through our regular audio system.
 
Remote control support: when the phone is on standby but something is playing in the background, instead of accessing the program to press pause/stop etc you just double tap the home button (the button located below the screen - is that it's name?) and you can remotely control what's playing in the background.

Sounds like you have it all sussed already. I can't wait to try out this feature.
 
Top