Start a new topic

Dragon Naturally Speaking *text-to-speech*

I've been reading that CTE works well with Dragon Naturally Speaking on Windows. I'm not interested in dictation, but in text-to-speech as a proofing method. 


Has any one tried it with the CTE's Autopilot function?


Thanks


Or, any other text-to-speech software other than Dragon?

Hello Mario, DNS will do a fine job, but the question is whether it works well in the CT editor. If I remember correctly, there were some issues. You will have to test it, however, even if not, DNS is fine for any other job on your PC.


Did you ever test it with Cortana (if there is an offline version, as there is for the equivalent in Mac)? 

Hello Tre. Before originating this thread I made a research and found that at least one CTE user is using or has used it satisfactorily with DNS (on Windows). 


I'm not interested in dictation, only text-to-speech, because I believe that text-to-speech combined with the Autopilot function can do a great proofreading job. There are many text-to-speech software around, but most work on the cloud, which is not desirable.

Hello Mario,


I'm not on Windows (as you know already), but I have heard good things about TextAloud, especially used in a proofreading scenario (the TTS engine reads the source text while you read the target text, or vice versa).


I don't think any TTS solution would play well with Autopilot, as it would have to be synced to the same characters limitations, plus have a means of reading each target segment as CafeTran presents it.


Can CafeTran copy automatically the target text to the clipboard, so that you can launch a TTS query? 


Also, I'm not sure if Autopilot can be used with Edit > Bind external editor.


For TTS purposes, instead of Autopilot, I would suggest you try the Project > Export and exchange > Export as bilingual document, which is also meant for (external) translation reviewing. For native projects, changes can even imported back to CafeTran, but you could use track changes to quickly review the changes to make for non-native projects, where reimporting the bilingual doc is not an option.


You can select the source or target column only and test the built-in Word TTS or any other TTS solution, which will hopefully allow you to customize the voice speed, pitch, etc.


Jean

Hello Jean,


Thanks you! While I still hope to hear from Igor about whether adding TTS to CTE one day is technically feasible (would it perhaps become the first TTS-capable CAT tool on the market?)—and other CTE's user opinion on this very matter—your suggestion to go the bilingual document is great.


I gave a look at TextAloud and listened to some Italian voice samples, which sounded similar to Word's TTS voices—not perfect but workable and good enough for proofreading. The price is very affordable too compared to Dragon Naturally Speaking, so I'll give it a try as soon as possible.


By the way, how do other CTE colleagues think of text-to-speech as a proofreading means? Have they tried it? With what results?

JeanI'm not on Windows...


I perfectly understand your decision to stick to Linux. (We tried Ubuntu several times, but we always went back to Windows after a few weeks. We also tried Mac when we were close to deciding to convert our Windows network to a Mac network, and still have one iMac which every now and then I look at with a mixed feeling.) 


But, besides being more difficult to learn, Linux in particular suffers too many software limitations for us translators. And while I'm continuously swearing that sooner or later our next attempt with Linux will be the final and the winning one on the way to abandoning Windows altogether, too often we realize that Windows is still an amazingly comfort zone. 


Linux takes a lot of courage, after all.

Mario: By the way, how do other CTE colleagues think of text-to-speech as a proofreading means? Have they tried it? With what results?


I find it a really interesting idea and I've been toying with it for some time now. I don't do much proofreading as I prefer translating, but I see the potential for self-revision.

TTS solutions do not abound on Linux. I have tested a script (Amazon Polly) that creates an mp3 file of any source or target txt. Lately, I'm testing the excellent "Read Aloud" Chrome extension, which provides access to voices from Amazon Polly, Google Wavenet, Microsoft, IBM Watson, all while allowing to change the speed, the pitch and the volume, and offering an option to highlight the readed text.

Exporting to TXT and saving as HTML does the trick, for now, but I'm just experimenting, no real production use so far.
---


Mario: Linux takes a lot of courage, after all.

 


I guess it does, but I would not have it any other way.


When I went freelance, I took up the challenge of working exclusively from my GNU/Linux OS. No buts, no dual boots…


I can now say it is entirely feasible, and enjoyable.


Having an excellent CAT tool such as CafeTran is a real boon for anyone looking to take the plunge. 


I have even devised a wiki to make the conversion simpler: https://translateonlinux.org/

JeanI don't do much proofreading as I prefer translating, but I see the potential for self-revision.


My interest too is confined to self-revision. Some translators prefer to proofread their work on print-outs, others prefer to have it read aloud.


Amazon Polly: This solution might be very interesting thanks to the API, which might make things easier for CTE besides being cross-platform.


Login to post a comment