aurum-core

Reusable on-device speech-to-text library for Aurum.

Audio in. Text out. On-device by default.

• Local whisper.cpp (whisper-rs, Metal on macOS)
• Optional OpenRouter (LLM-assisted ASR)
• PCM-first mic host API · partial-window helpers · cancel
• Cleanup / flow (rules or OpenRouter)
• Model download, pins, txt/srt/json

API status: experimental until 0.1.0. Pin a git rev or tag.

Depend

aurum-core = { git = "https://github.com/joe-broadhead/aurum", package = "aurum-core", tag = "v0.0.0" }
# or: path = "../aurum/crates/aurum-core"
tokio = { version = "1", features = ["rt-multi-thread", "macros"] }

Quick example

use aurum_core::audio::load_audio;
use aurum_core::providers::{LocalWhisperProvider, TranscriptionOptions};

# async fn demo() -> aurum_core::Result<()> {
let audio = load_audio(std::path::Path::new("meeting.m4a")).await?;
let provider = LocalWhisperProvider::new(std::path::PathBuf::from("/tmp/aurum-cache"))
    .with_progress(false);
let result = provider
    .transcribe(
        &audio,
        &TranscriptionOptions {
            model: "tiny-q5_1".into(),
            language: "en".into(),
            timestamps: true,
            cancel: None,
        },
    )
    .await?;
println!("{}", result.text);
aurum_core::providers::local::clear_context_cache();
# Ok(())
# }

Docs: https://joe-broadhead.github.io/aurum/library/integration/

Build requirements

• Build: cmake + C/C++ toolchain
• Runtime (files): ffmpeg for non-16 kHz mono WAV

License

MIT