Options to configure the MediaPipe Audio Transcriber Task.
Signature:
export declare interface AudioTranscriberOptions extends TaskRunnerOptions
Extends: TaskRunnerOptions
Properties
| Property | Type | Description |
|---|---|---|
| cacheDir | string | undefined | Optional directory path for caching compiled XNNPACK / WebGPU weights in the WASM virtual filesystem. Defaults to '/tmp'. |
| enableInterimResults | boolean | undefined | Whether to emit non-final (interim) transcription segments during streaming transcription. Defaults to false. |
| modelArchitecture | ModelArchitecture | number | undefined | The ASR model architecture. Defaults to 'AUTO'. |
| modelAssetSize | number | undefined | Total byte size of the model asset when streaming via ReadableStreamDefaultReader. |
| prompt | string | undefined | Optional prompt / instruction prefix for decoder-based ASR architectures (such as 'TINY_GEMMA'). |
| sampleRateHz | number | undefined | Sample rate of the input audio in Hz. Defaults to 16000. |
AudioTranscriberOptions.cacheDir
Optional directory path for caching compiled XNNPACK / WebGPU weights in the WASM virtual filesystem. Defaults to '/tmp'.
Signature:
cacheDir?: string;
AudioTranscriberOptions.enableInterimResults
Whether to emit non-final (interim) transcription segments during streaming transcription. Defaults to false.
Signature:
enableInterimResults?: boolean;
AudioTranscriberOptions.modelArchitecture
The ASR model architecture. Defaults to 'AUTO'.
Signature:
modelArchitecture?: ModelArchitecture | number;
AudioTranscriberOptions.modelAssetSize
Total byte size of the model asset when streaming via ReadableStreamDefaultReader.
Signature:
modelAssetSize?: number;
AudioTranscriberOptions.prompt
Optional prompt / instruction prefix for decoder-based ASR architectures (such as 'TINY_GEMMA').
Signature:
prompt?: string;
AudioTranscriberOptions.sampleRateHz
Sample rate of the input audio in Hz. Defaults to 16000.
Signature:
sampleRateHz?: number;