This module provides core functionality for testing library initialization and detecting hardware acceleration support on different platforms.
testSherpaInit() Test method to verify that the sherpa-onnx native library is loaded correctly.
function testSherpaInit () : Promise < string >
Returns Resolves with a test message confirming the library is loaded.
Example import { testSherpaInit } from 'react-native-sherpa-onnx' ; const result = await testSherpaInit (); console . log ( result ); // "Sherpa ONNX initialized successfully"
getQnnSupport() Check Qualcomm QNN (Qualcomm Neural Network) acceleration support on Android devices.
function getQnnSupport ( modelBase64 ?: string ) : Promise < AccelerationSupport >
Parameters Optional base64-encoded model for session initialization test. If omitted, uses an embedded test model.
Returns Whether QNN provider is compiled into the library
Whether a QNN-compatible hardware accelerator is available
Whether a model can be initialized using QNN
Android : Full support on Qualcomm devicesiOS : Returns all false
Example import { getQnnSupport } from 'react-native-sherpa-onnx' ; const qnnSupport = await getQnnSupport (); if ( qnnSupport . canInit ) { console . log ( 'QNN acceleration available' ); // Use provider: 'qnn' in model options }
getNnapiSupport() Check NNAPI (Android Neural Networks API) acceleration support on Android devices.
function getNnapiSupport ( modelBase64 ?: string ) : Promise < AccelerationSupport >
Parameters Optional base64-encoded model for session initialization test.
Returns
Android : Support varies by device and Android versioniOS : Returns all false
Example import { getNnapiSupport } from 'react-native-sherpa-onnx' ; const nnapiSupport = await getNnapiSupport (); if ( nnapiSupport . canInit ) { // Use provider: 'nnapi' in model options }
getXnnpackSupport() Check XNNPACK (CPU-optimized) acceleration support.
function getXnnpackSupport ( modelBase64 ?: string ) : Promise < AccelerationSupport >
Parameters Optional base64-encoded model for session initialization test.
Returns Whether XNNPACK provider is compiled into the library
Returns true when providerCompiled is true (CPU-optimized execution)
Whether a model can be initialized using XNNPACK
Android : Full support (CPU-optimized inference)iOS : Returns all false
Example import { getXnnpackSupport } from 'react-native-sherpa-onnx' ; const xnnpackSupport = await getXnnpackSupport (); if ( xnnpackSupport . canInit ) { // Use provider: 'xnnpack' in model options }
getCoreMlSupport() Check Core ML acceleration support on iOS devices with Apple Neural Engine.
function getCoreMlSupport ( modelBase64 ?: string ) : Promise < AccelerationSupport >
Parameters Optional base64-encoded model for session initialization test.
Returns True on iOS 11+ (Core ML framework available)
True when Apple Neural Engine is available (A11+ chips)
Whether a model can be initialized using Core ML
iOS : Full support on iOS 11+ with Apple Neural EngineAndroid : Returns all false
Example import { getCoreMlSupport } from 'react-native-sherpa-onnx' ; const coreMLSupport = await getCoreMlSupport (); if ( coreMLSupport . hasAccelerator ) { console . log ( 'Apple Neural Engine available' ); // Use provider: 'coreml' in model options }
getAvailableProviders() Get the list of available ONNX Runtime execution providers on the current device.
function getAvailableProviders () : Promise < string []>
Returns Array of provider names (e.g., ["CPU", "NNAPI", "QNN", "XNNPACK"]).
Requires the ONNX Runtime Java bridge from the onnxruntime AAR.
Example import { getAvailableProviders } from 'react-native-sherpa-onnx' ; const providers = await getAvailableProviders (); console . log ( 'Available providers:' , providers ); // ["CPU", "XNNPACK", "QNN"] on Qualcomm Android // ["CPU", "CoreML"] on iOS with Neural Engine
Types AccelerationSupport Result type for hardware acceleration queries.
interface AccelerationSupport { /** Whether the provider is compiled into the library */ providerCompiled : boolean ; /** Whether compatible hardware accelerator is available */ hasAccelerator : boolean ; /** Whether a model session can be initialized with this provider */ canInit : boolean ; }
See Also 