Crate eredu_codec 

Backend-neutral neural audio codec architectures.

This crate keeps codec implementations optional and separate from eredu. Realtime language models can operate on discrete codec tokens, while applications that need audio encode/decode can depend on this crate.

Modules

mimi

Mimi neural audio tokenizer support. Mimi neural audio tokenizer support.

Structs

AudioTokenizerConfig

Static metadata for pairing an audio tokenizer with a realtime model.

Enums

Error

Errors returned by codec loaders and tokenization operations.

Traits

AudioTokenizer

Common interface for neural audio tokenizers.