Closed jjbosox closed 8 years ago
I think it's a valuable addition to Streams - was thinking in this direction, but with Sphinx. Kaldi as I understand provides no ready solution, but only tools to build one. What is the basic functionality you're going to introduce? Is it only English? Any text or fixed dictionary? +1
This toolkit will also provide operators to help with load-management
streamsx.plumbing has some load management operators. If these operators are not specific to speech contributing them to streamsx.plumbing would be appreciated.
+1 Will Watson integration include Watson services on Bluemix?
+1 Agree with @ddebrunner on load management operators being in streamsx.plumbing if they are to specific to speech.
@leongor
I see no reason we could not include Sphinx it would simply be a matter of priorities . Kaldi does provide a nice ready solution but only for English. A sleu of recipes exist such that other ready solutions are also available provided you have the corpus to work from .
@ddebrunner
i will take a look at streamsx.plumbing . currently the operator mentioned is not generic in nature but its worth thinking about .
At some point yes it would be well worth having integration with the Watson Services. our priorities right now however are to integrate with the code directly
+1
+1
Please advise on who I should put as committers to the project. Please sign and mail a copy of this agreement to me and @petenicholls : https://github.com/IBMStreams/administration/blob/master/IBMStreams-cla-individual.pdf
I will create the repository once I receive the signed agreement.
Thanks!
@jjbosox , please advise who the committers are, and if you have signed the CLA, so I can create the repository.
Thanks!
created streamsx.speech2text.
Initial committer is @jjbosox
I am proposing creation of a speech-to-text toolkit.
Speech-to-text processing fits well into the Streams paradigm. Call-centers around the word are trying to leverage analytics on their phone calls ( audio ) to provide better service at a lower cost. This toolkit will provide integration with commonly used speech-to-text engines. This toolkit will also provide operators to help with load-management because of the performance intensity of speech-to-text translation. As a starting point, we will provide integration with Kaldi and Watson. Users will need to have Watson and/or Kaldi installed on the systems to get the sample applications and operators to function properly. This toolkit will not include IBM Watson IP.