jfainberg/lattice_combination

Lattice combination algorithm to combine inaccurate transcripts with hypothesis lattices

37
/ 100
Emerging

This project helps speech recognition researchers improve the accuracy of their acoustic models by combining imperfect speech transcripts with "hypothesis lattices" – a graphical representation of possible spoken words. It takes in existing, error-prone transcripts and generated speech lattices, producing refined lattices that can be used to train more robust acoustic models. This is for researchers and engineers working on speech-to-text systems, particularly those with limited perfectly transcribed audio data.

No commits in the last 6 months.

Use this if you are training acoustic models for speech recognition and need to make the best use of a mix of high-quality and somewhat inaccurate speech transcripts.

Not ideal if you are not working with Kaldi-based speech recognition systems or if you have ample, perfectly transcribed audio for your training data.

speech-to-text acoustic-modeling speech-recognition semi-supervised-learning natural-language-processing
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 6 / 25
Maturity 16 / 25
Community 15 / 25

How are scores calculated?

Stars

16

Forks

5

Language

Jupyter Notebook

License

Apache-2.0

Last pushed

Mar 19, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/voice-ai/jfainberg/lattice_combination"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.