cjpais/Handy · error · anyhow::Error

Failed to load parakeet model

Error message

Failed to load parakeet model {}: {}

What it means

Raised when ParakeetModel::load(path, Quantization::Int8) fails to load an ONNX-based Parakeet model via transcribe-rs. Failures come from the ONNX runtime layer: unreadable/corrupt .onnx files, missing companion files for the model directory, unsupported opset, or execution-provider init failure. A 'loading_failed' event is emitted with the same message before the error is returned.

Solutions

1. Re-download the Parakeet model from the catalog (downloads are hash-verified, guaranteeing intact files).
2. Restart the app after changing accelerator settings so a fresh ONNX session/provider is initialized.
3. Try CPU acceleration to rule out a missing execution provider.
4. If it persists after a clean re-download, the model's opset may exceed the bundled runtime — report the model id and app version.

// Verify all expected ONNX files exist before loading
for f in ["model.onnx", "vocab.json"] {
    if !model_path.join(f).exists() {
        anyhow::bail!("parakeet model incomplete: missing {} in {}", f, model_path.display());
    }
}

match ParakeetModel::load(&model_path, &Quantization::Int8) {
    Ok(m) => Ok(m),
    Err(e) => {
        warn!("parakeet Int8 failed ({e}); retrying default quantization on CPU");
        set_accelerator_cpu();
        ParakeetModel::load(&model_path, &Quantization::default())
    }
}

When it happens

Trigger: Selecting a Parakeet model whose ONNX files are corrupt or incompletely downloaded; opset/operators unsupported by the bundled onnxruntime; provider (e.g. CUDA/CoreML) unavailable on the machine; quantized graph incompatible with the runtime build.

Common situations: Interrupted model downloads leaving truncated .onnx files; onnxruntime version drift after dependency updates; machines without the expected acceleration providers; models added by hand into the models directory without all required files.

Related errors

• Failed to create SileroVad
• Failed to load canary model
• Failed to load cohere model
• Failed to load gigaam model
• Failed to load moonshine model

AI-assisted analysis of cjpais/Handy@ba10ce1943 (2026-08-17). Data as JSON: /api/errors/ddb87f221dadc76e. Report an issue: GitHub.

Appendix: source

Thrown at src-tauri/src/managers/transcription.rs:645

                     supports_translate={}, supports_language_detect={})",
                    model_id,
                    backend,
                    requested_device,
                    bound_backend,
                    bound_device,
                    caps.supports_streaming,
                    caps.supports_translate,
                    caps.supports_language_detect
                );
                LoadedEngine::TranscribeCpp(session)
            }
            EngineType::Parakeet => {
                let engine =
                    ParakeetModel::load(&model_path, &Quantization::Int8).map_err(|e| {
                        let error_msg =
                            format!("Failed to load parakeet model {}: {}", model_id, e);
                        emit_loading_failed(&error_msg);
                        anyhow::anyhow!(error_msg)
                    })?;
                LoadedEngine::Parakeet(engine)
            }
            EngineType::Moonshine => {
                let engine = MoonshineModel::load(
                    &model_path,
                    MoonshineVariant::Base,
                    &Quantization::default(),
                )
                .map_err(|e| {
                    let error_msg = format!("Failed to load moonshine model {}: {}", model_id, e);
                    emit_loading_failed(&error_msg);
                    anyhow::anyhow!(error_msg)
                })?;
                LoadedEngine::Moonshine(engine)
            }
            EngineType::MoonshineStreaming => {
                let engine = StreamingModel::load(&model_path, 0, &Quantization::default())

View on GitHub (pinned to ba10ce1943)