https://github.com/dialogflow/api-ai-english-asr-model
Api.ai English Speech Recognition (ASR) Model for Kaldi
https://github.com/dialogflow/api-ai-english-asr-model
Last synced: 4 months ago
JSON representation
Api.ai English Speech Recognition (ASR) Model for Kaldi
- Host: GitHub
- URL: https://github.com/dialogflow/api-ai-english-asr-model
- Owner: dialogflow
- License: apache-2.0
- Created: 2016-03-30T23:13:43.000Z (about 9 years ago)
- Default Branch: master
- Last Pushed: 2020-12-27T12:21:27.000Z (over 4 years ago)
- Last Synced: 2025-01-11T02:13:49.563Z (6 months ago)
- Homepage:
- Size: 5.86 KB
- Stars: 36
- Watchers: 15
- Forks: 10
- Open Issues: 3
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Api.ai kaldi Speech Recognition Model
Model is released under Creative Commons Attribution-ShareAlike 4.0 International Public License and available here for download here https://api.ai/downloads/api.ai-kaldi-asr-model.zip
## How to use
There are example scripts on how to use model with model in kaldi repo
https://github.com/kaldi-asr/kaldi/blob/master/egs/apiai_decode/s5/README
You can also test this model using your borwser and HTTP protocol, follow guide here:
https://github.com/api-ai/asr-server
## Training
Model is trained using kaldi "Chain models", read more about them here http://kaldi-asr.org/doc2/chain.html
Acustic data for training is Api.ai data mixed with Librispeech corpus (http://www.openslr.org/12/)
Language model is based on Assistant.ai logs, so it good for short commands, like "Wake me up at 7 am".
## Results
Api.ai test corpus WER: 11.2%