Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

simple wav decoding don't work #30

Open
by-tech opened this issue Dec 1, 2017 · 3 comments
Open

simple wav decoding don't work #30

by-tech opened this issue Dec 1, 2017 · 3 comments

Comments

@by-tech
Copy link

by-tech commented Dec 1, 2017

En_1272-128104-0000.zip
installation and compile was success. But when trying a simple wav english file, there's no recognition.

to be sure that file is correct , i used "ffmpeg -i En_1272-128104-0000.wav -f s16le -ar 16000 -ac 1 En_1272-128104-0000.raw"

what i have missing ?
thanks to help

RESULT

curl -H "Content-Type: application/octet-stream" --data-binary En_1272-128104-0000.raw http://localhost/asr

{"status":"ok","data":[{"confidence":0.932974,"text":""}]}

VLOG[4] (fcgi-nnet3-decoder[5.2.215~1-5e7d]:FinalizeDecoding():lattice-faster-online-decoder.cc:788) pruned tokens from 4053 to 74
VLOG[4] (fcgi-nnet3-decoder[5.2.215~1-5e7d]:GetRawLattice():lattice-faster-online-decoder.cc:191) init:40 buckets:83 load:0.891566 max:1
VLOG[1] (fcgi-nnet3-decoder[5.2.215~1-5e7d]:DeterminizeLatticePhonePruned():determinize-lattice-pruned.cc:1440) Doing first pass of determinization on phone + word lattices.
VLOG[1] (fcgi-nnet3-decoder[5.2.215~1-5e7d]:DeterminizeLatticePhonePruned():determinize-lattice-pruned.cc:1455) Doing second pass of determinization on word lattices.
VLOG[1] (fcgi-nnet3-decoder[5.2.215~1-5e7d]:Decode():OnlineDecoder.cc:250) Recognized @ 71 ms
VLOG[1] (fcgi-nnet3-decoder[5.2.215~1-5e7d]:Decode():OnlineDecoder.cc:255) Decode subroutine done
@viju2008
Copy link

I am also facing same issue

@mikenewman1
Copy link

Me too #31
What version of Kaldi are you using? Something recent?
And what version did you use to build your models?
It works for me if I use a model from about a year ago.

@bjascob
Copy link
Contributor

bjascob commented Jan 5, 2018

Back in #11 I posted a tar of some test code (see https://github.com/api-ai/asr-server/files/828349/APIAI_Server.tar.gz) that allows you to input xx.raw files directly. I just tried that code again today against Kaldi pulled from Github today and it worked for me (note that to compile you need to remove kaldi-thread.a from src/Makefile). The output for En_1272-128104-0000.raw was...

{"status":"ok","data":[{"confidence":0.897815,"text":"MR QUARTERS THE APOSTLE OF MIDDLE CLASS AND WE'RE GLAD WELCOME HIS GOSPEL"}]}

You might try that code. It's a bit hackerish but if you compile and execute run.sh it'll decode the included test.raw. Be sure to change the location of the model as defined at the top of run.sh.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants