Florian Metze
can you provide more information? a more complete log file? which OS, libraries, etc?
in the tensorflow framework, this should be very easy. in the c++ framework, i don't think there is an easy way to do this.
Can you be a bit more specific which code you are using for each of the steps, and which results you are seeing? This is kind of hard to debug!...
not sure - the last activity was @ramonsanabria who said "testing this now"?
Yes, right now, we are only using one pronunciation, whichever one happens to be last (I believe). This is clearly not the best thing we can do, but dictionary learning...
No, I don’t think so. > On Nov 28, 2017, at 5:28 PM, Jinserk Baik wrote: > > Thank you for replying, @riebling . Then do we have to assume...
yeah, what i am saying is that BLSTM CTC models work remarkably well for a dictionary that consists simply of the English characters. It is generally not better than a...
Not sure I understand. The whole point of a character based network is that if you want to add a new word, you know what the representation will be. No...
yes, this is exactly what is happening. the character AM is learning “typical" G2P rules. > On Nov 29, 2017, at 12:13 PM, Jinserk Baik wrote: > > Ah, I...
Apologies for the late response, I'd been traveling a lot. I think it should be possible to generate a "fake" model file that corresponds to the Eesen context independent acoustic...