Skip to main content
Interface for configuring the Whisper context during initialization with initWhisper(). Controls model loading, GPU acceleration, and Core ML settings.

​
Properties

​
filePath
string | number
required
Path to the GGML model file or asset require number.Can be:
  • Absolute file path: '/path/to/model.bin'
  • Asset require: require('./assets/model.bin')
  • Bundle asset name (iOS/Android): 'model.bin' with isBundleAsset: true
​
coreMLModelAsset
object
Core ML model assets configuration. Required when using require() for the model path and enabling Core ML on iOS.
​
isBundleAsset
boolean
Whether the file path is a bundle asset name (for pure string filePath). Set to true if the model is bundled in the app’s assets.
​
useCoreMLIos
boolean
default:"true"
Prefer to use Core ML model if it exists (iOS only). If set to false, even if the Core ML model exists, it will not be used.Core ML accelerates the encoder on iOS 15.0+.
​
useGpu
boolean
default:"true"
Use GPU/Metal acceleration if available (iOS only). When enabled, uses Metal for GPU-accelerated inference.If enabled, Core ML option will be ignored.
​
useFlashAttn
boolean
default:"false"
Use Flash Attention optimization. Only recommended if GPU is available. Can significantly improve performance on supported devices.

​
Usage Examples

​
Basic Initialization

import { initWhisper } from 'whisper.rn'

const context = await initWhisper({
  filePath: '/path/to/ggml-tiny.en.bin'
})

​
Using Asset Require

const context = await initWhisper({
  filePath: require('../assets/ggml-tiny.en.bin')
})

​
With GPU Acceleration

const context = await initWhisper({
  filePath: '/path/to/ggml-base.bin',
  useGpu: true,
  useFlashAttn: true
})

​
With Core ML (iOS)

const context = await initWhisper({
  filePath: require('../assets/ggml-tiny.en.bin'),
  useCoreMLIos: true,
  coreMLModelAsset: {
    filename: 'ggml-tiny.en-encoder',
    assets: [
      require('../assets/ggml-tiny.en-encoder/model.mil'),
      require('../assets/ggml-tiny.en-encoder/coremldata.bin'),
      require('../assets/ggml-tiny.en-encoder/weights/weight.bin'),
      require('../assets/ggml-tiny.en-encoder/analytics/coremldata.bin'),
    ]
  }
})

​
Bundle Asset (iOS/Android)

const context = await initWhisper({
  filePath: 'ggml-tiny.en.bin',
  isBundleAsset: true
})

​
Platform Notes

​
iOS

  • Core ML models must be co-located with the GGML model
  • Core ML model directory format: <model-name>-encoder.mlmodelc/
  • GPU acceleration uses Metal framework
  • Flash Attention requires GPU to be enabled

​
Android

  • GPU acceleration is not currently supported
  • Core ML options are ignored on Android
  • Use isBundleAsset: true for assets in android/app/src/main/assets/

​
Checking GPU Status

const context = await initWhisper({
  filePath: '/path/to/model.bin',
  useGpu: true
})

console.log('GPU enabled:', context.gpu)
if (!context.gpu) {
  console.log('Reason:', context.reasonNoGPU)
}

Build docs developers (and LLMs) love

Get started for freeTalk to us