Extending Gentle Aligner
Week 3
Visualization using Javascript
- Step 1: Data preparation, Kaldi format to JSON, Generating JSON scripts for word and phoneme alignment files
timed_decoding/ - index.html
-data/
- wav/
* [sample_name].wav
* …
- json/
* meta.json
#Should contain an object with the following mapping:
{sample_name: <str>,
ground_truth: <transcript string>}, …}
* [sample_name].words.json
[{text: <str>,
start_time: <float>,
end_time: <float> }, …]
* [sample_name].phonemes.json
[{text: <str>,
start_time: <float>,
end_time: <float> }, …]
-
Step 2: Creating a react app
-
Step 3: Deploying the react app on jekyll blog
yarn run build cp -r build blog-path/assets/ -
Visualization of generated alignments using javascript
- Next: Automating Decoding Process Week 4
- Prev: Decoding Voxforge_ru Russian ASR model using recipe Week 2
- main page
Tools: Kaldi, Python, C, Bash Scripting
Link to GSoC Project Repository