asr_eval¶
Evaluation and building components for Automatic Speech Recognition (ASR) in Python.
See the preprint paper and full PDF documentation.
Evaluating ASR with extended annotation syntax
Running and evaluating non-streaming ASR, and perform a detailed error analysis
Running and evaluating streaming ASR, producing many informative diagrams
Alignment features
Streaming features
Installation
Alignments and WER
Streaming evaluation
Evaluation and dashboard
Models list
Datasets list
Acknowledgements¶
Several components of this project, related to the correction module, were developed in
collaboration with Yandex Camp students during July 2025:
Timur Rafikov,
Vasily Kudryavtsev,
Yana Fitkovskaja,
Dmitry Ezhov,
Nikita Vambrikov,
Anastasia Zvereva,
Valeria Stolyarova,
Mukharyam Baviev,
Antal Schiff,
Mirzomansurkhon Sultanov.