Comments (4)
The prediction model is similar to a language model, its actually phone language model for TIMIT dataset.
First add <SOS>
<EOS>
to each transcription, here I would like to use blank
label as both <SOS>
and <EOS>
.
The input of prediction model is a sequence of phones with blank
label prepended, the output should be the corresponding phone sequence with blank
appended.
In one word, the training stage is same to language model, except it has a blank
label.
from rnn-transducer.
Thanks for your prompt response. Just a follow up question, if I were to train character-wise LM, for instance: for a sentence like "hello world", should I send the following as input:
h e l l o w o r l d
and the following as target:
h e l l o w o r l d
I really appreciate your answer!
from rnn-transducer.
Well, for rnn-transducer, like CTC, it introduced a blank
label, so you need to send the following as input (_
means blank
) :
_ h e l l o w o r l d
and the following as the target:
h e l l o w o r l d _
from rnn-transducer.
Somehow the blank tag got dropped in my post :(
Yes, I meant by your response. Thanks a lot!
from rnn-transducer.
Related Issues (20)
- greedy_decorder no output HOT 4
- mxnet.base.MXNetError: [11:42:46] src/operator/contrib/./rnnt_loss-inl.h:218: Check failed: dshape[0] == lshape[0] (474 vs. 1) The batch size for the labels and data arrays must be the same. HOT 16
- Put the testing process to gpu HOT 6
- Why adding a while loop when decoding on the gpu? HOT 10
- The performance gets worse when I use a bigger model HOT 11
- Phoneme Mapping HOT 2
- RNNTLoss issue with model 2013 HOT 2
- MXNet with Version 1.1.0 doesn't contain the install files (/docs/install), Version 1.4.x doesn't work (Shape Error) HOT 1
- Why should we use log_softmax when training?
- Loss is decrease but SER is increase
- A question about vocab_size
- mxnet.base.MXNetError: [16:48:44] ../src/operator/./rnn-inl.h:507: RNN on GPU is only available for cuDNN at the moment. HOT 1
- A Question About Padding
- What is the license for the code/model in this repository?
- DataLoader.py return these errors... HOT 1
- Model 2012 reproduce HOT 2
- A problem about getting 123 dim feature
- question in greedy decoder HOT 2
- Dataset and run.sh HOT 12
- src/operator/rnn.cc:254:30: error: ‘kCuDNNDropoutDesc’ is not a member of ‘mxnet::ResourceRequest’ HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from rnn-transducer.