Does Kaldi Support Idempotent Write Retries Internally

Kaldi Data Integrity

The goal is to ensure that repeated write operations to output artifacts do not result in data duplication within the Kaldi pipeline.

Constraints

Kaldi's core functionality focuses on acoustic model training and decoding, not transactional data management.

When managing external write retries, what specific configuration parameters govern output file generation?

• Does the toolkit offer a built-in mechanism for detecting prior successful writes?
• What is the documented behavior when an output file already exists during a pipeline run?