- This topic has 10 replies, 2 voices, and was last updated 9 years, 5 months ago by Halle Winkler.
-
AuthorPosts
-
June 5, 2012 at 9:51 pm #9715mariagrinevaParticipant
I am trying to use OpenEars as an additional tool to Nuance that would help it with recognizing names from person’s contact list.
For example, I would send to Nuance “Call my friend Maxim” and usually it hears “mexican” instead. And I would have a separate OpenEars thread that would specifically listen to friend’s names only. And if this thread could catch “Maxim” in my phrase, that would be fantastically helpful to me!
Now, I have tried to create a custom dictionary made of friends’ names and run OpenEars sample. And it recognized the whole phrase “Call my friend Maxim” into words of this custom dictionary. It has caught ‘Maxim’ correctly, but also it has drew up other words to my friends’ names. Is it possible to have a separate recognitionScore for each recognized word in the input?
Thanks
June 5, 2012 at 10:05 pm #9716Halle WinklerPolitepixOK, to clarify my own understanding of the issue, is what is happening that the actual phrase is “call my friend Maxim” and what is being reported is “Molly Glen Maxim” or something along those lines with the non-name words being replaced by similar-sounding names? Or is the issue that the rest of the sentence is being recognized correctly but you want to disregard non-name words which are in your language model?
The end goal is just to be informed that “Maxim” was detected in the sentence without needing to know the specifics about the other words, is that correct? I don’t think there is a way to get per-word scores for a multiple word sentence, but n-best scoring will be coming up in the next version of OpenEars.
June 5, 2012 at 11:32 pm #9717mariagrinevaParticipant“the actual phrase is “call my friend Maxim” and what is being reported is “Molly Glen Maxim” or something along those lines with the non-name words being replaced by similar-sounding names?” yes, exactly.
Similarly, if I say ‘Call restaurant’ it would recognize ‘Molly Cris Ann’
I am looking for a way to know that ‘Maxim’ In the first case was recognized with a higher probabilityJune 6, 2012 at 4:20 pm #9718Halle WinklerPolitepixOK, I think that the upcoming n-best feature should handle this for you but there isn’t anything in the current version which I can think of which will help. I’m hoping to release the next version around next Monday.
June 6, 2012 at 4:28 pm #9719mariagrinevaParticipantHalle, next Monday would be great! Looking forward to try out the idea. Please let me know if I can get the next cersionearlier to play with it, even if it’s still buggy
June 16, 2012 at 8:10 pm #9781Halle WinklerPolitepixHi Maria,
The new version is up (I haven’t had time to update the detailed documentation so I haven’t announced it yet but you can see it at https://www.politepix.com/openears) so you can give it a try. There is a preprocessor define that turns on n-best in the new sample app so you can uncomment it to experiment with n-best and scoring.
July 12, 2012 at 11:57 pm #10538mariagrinevaParticipantHi Halle,
Thanks for the update! What should I uncomment to try the new feature?
July 13, 2012 at 10:25 am #10539Halle WinklerPolitepixHi Maria,
Take a look at the sample app and search for “nbest”.
July 13, 2012 at 9:00 pm #10540mariagrinevaParticipantHi Halle,
Thanks, I figured it out and ran the sample. Yeah, n-best is not exactly what I need (I need rather per-word score), but will try to play with it and see.
Another question: how big the grammar can be? I have a vocabulary of around 2000 words, is it OK?
July 13, 2012 at 9:16 pm #10541Halle WinklerPolitepixSounds a little big for local recognition (I think 500-1000 words is probably better) but the only way to know for sure is to test.
Try setting n-best to return the 1 best hyp with a score and I think you should receive the score per word.
April 27, 2014 at 4:50 pm #1021049Halle WinklerPolitepixThis concept might work well with the new API for dynamically generating rules-based grammars with OpenEars.
-
AuthorPosts
- You must be logged in to reply to this topic.