githubharald/CTCDecoder

Connectionist Temporal Classification (CTC) decoding algorithms: best path, beam search, lexicon search, prefix search, and token passing. Implemented in Python.

61
/ 100
Established

This tool helps developers working with Connectionist Temporal Classification (CTC) neural networks interpret their model's raw output. You feed it the confidence scores (a matrix) from your CTC-trained neural network and a list of possible characters, and it returns the most probable text string. It is used by machine learning engineers or researchers building systems for tasks like speech recognition or handwriting recognition.

835 stars.

Use this if you need to convert the probabilistic output of a CTC-trained neural network into a readable sequence of characters or words, with options to improve accuracy using language models or dictionaries.

Not ideal if you need a pre-built integration with TensorFlow or PyTorch, or if you are not working with CTC-trained neural networks.

speech-to-text handwriting-recognition optical-character-recognition sequence-prediction neural-networks
No Package No Dependents
Maintenance 10 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 25 / 25

How are scores calculated?

Stars

835

Forks

179

Language

Python

License

MIT

Last pushed

Jan 31, 2026

Commits (30d)

0

Get this data via API

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

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