react-native-executorch/src/types/vad.ts

An easy way to run AI models in React Native with ExecuTorch

import { ResourceSource } from '../types/common';
import { RnExecutorchError } from '../errors/errorUtils';

/**
 * Union of all built-in VAD model names.
 * @category Types
 */
export type VADModelName = 'fsmn-vad';

/**
 * Props for the useVAD hook.
 * @category Types
 * @property {object} model - An object containing the model configuration.
 * @property {VADModelName} model.modelName - Unique name identifying the model.
 * @property {ResourceSource} model.modelSource - The source of the VAD model binary.
 * @property {boolean} [preventLoad] - Boolean that can prevent automatic model loading (and downloading the data if you load it for the first time) after running the hook.
 */
export interface VADProps {
  model: { modelName: VADModelName; modelSource: ResourceSource };
  preventLoad?: boolean;
}

/**
 * Represents a detected audio segment with start and end timestamps.
 * @category Types
 * @property {number} start - Start time of the segment in seconds.
 * @property {number} end - End time of the segment in seconds.
 */
export interface Segment {
  start: number;
  end: number;
}

/**
 * React hook state and methods for managing a Voice Activity Detection (VAD) model instance.
 * @category Types
 */
export interface VADType {
  /**
   * Contains the error message if the VAD model failed to load or during processing.
   */
  error: null | RnExecutorchError;

  /**
   * Indicates whether the VAD model has successfully loaded and is ready for inference.
   */
  isReady: boolean;

  /**
   * Indicates whether the model is currently processing an inference.
   */
  isGenerating: boolean;

  /**
   * Represents the download progress as a value between 0 and 1.
   */
  downloadProgress: number;

  /**
   * Runs the Voice Activity Detection model on the provided audio waveform.
   * @param waveform - The input audio waveform array.
   * @returns A promise resolving to an array of detected audio segments (e.g., timestamps for speech).
   * @throws {RnExecutorchError} If the model is not loaded or is currently processing another request.
   */
  forward(waveform: Float32Array): Promise<Segment[]>;
}