Comments (3)
from icefall.
I notice that the function "intersect" of class "Nbest" calls k2.shortest_path which return the best path of the lattice.
Lines 765 to 767 in 6e43905
The
lattice
in this function is a FsaVec, which contains as many FSAs as the number of utterances. Note that each FSA in lattice
is NOT linear. Its scores
consists of acoustic scores (from the neural network) and LM scores (from the n-gram LM).
Line 820 in 6e43905
nbest
here conains an FsaVec. Each FSA in it is linear. Each utterance can have at most num_paths
linear FSAs.Note that at this point, the
scores
of each FSA is 0. When a linear FSA is intersected with another non-linear FSA, i.e., a lattice
, the resulting FSA is not linear and it contains all possible paths that are equivalent to the linear FSA. We use the shortest path to assign scores
to the original linear FSA (Remember that its initial scores
is 0). The sole purpose of k2.shortest_path
here is to assign scores to the original FSAs in the nbest
.from icefall.
Oh, I got it. Thank you so much. @danpovey @csukuangfj
from icefall.
Related Issues (20)
- Use CutSet.mux to effect? HOT 10
- Help with training/finetuning a zipformer based model HOT 6
- Different Training Loss with Single Node (8 GPUs) vs. Two Nodes (4 GPUs Each)
- Data cleaning HOT 3
- ONNX decode error HOT 2
- OTC with conformer librispeech/WASR isn't converage.
- ONNX bug HOT 9
- Questions about modifying prepare.sh for training ASR model on custom data HOT 2
- How to use my own dataset based on another dataset HOT 3
- kaldifeat installation error HOT 2
- Why unique lexicon is needed in Chinese ASR, but not in English ASR?
- Error during training OTC conformer_ctc2 HOT 1
- What is difference between zipformer and zipformer_ctc models? HOT 1
- ONNX and Torch models HOT 1
- how to decrease the right chunk size when using zipformer model?
- How to load the base model in the fine-tuning task of KWS HOT 3
- Getting segmentation fault HOT 9
- Training break down. details is showed below HOT 3
- training very slowly HOT 16
- Using my own data to train pruned_transducer_stateless7_ctc_bs, encountered an error
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 icefall.