Crate voxora_config 

Single source of truth for every voxora runtime setting.

Replaces the ad-hoc cascade currently duplicated across voxora-cli/src/args.rs::resolve_hf_cache_dir and voxora-hf/src/cache.rs::default_cache_root.

Settings come from three layers — built-in defaults, an optional TOML file, and the environment — and every getter resolves them in the same order, first non-empty wins:

1. An explicit value on the config itself: either set by the caller (CacheConfig::root, HfConfig::token, …) or loaded from a TOML file via VoxoraConfig::from_file.
2. The matching VOXORA_* environment variable, listed in env (plus the HF_TOKEN / HUGGING_FACE_HUB_TOKEN aliases for the token).
3. The built-in default (VoxoraConfig::default).

ASR-specific: this crate only models configuration concerns that exist in the voxora speech-recognition stack (cache dir, HF token, HF base URL, default revision). It does not model LLM / vision / multimodal configuration.

Example

use voxora_config::VoxoraConfig;

let cfg = VoxoraConfig::from_file(std::path::Path::new("voxora.toml"))?;
println!("cache root: {}", cfg.cache_root().display());
println!("hub: {}", cfg.hf_base_url());

Modules

env

Centralised list of VOXORA_* environment variables.

file

Optional TOML file loader for voxora-config.

Structs

CacheConfig

Where voxora keeps downloaded models on disk.

HfConfig

How voxora talks to the Hugging Face Hub.

VoxoraConfig

Top-level voxora configuration.

Enums

ConfigError

All errors that may occur while loading a voxora configuration.