Cheetah Speech-to-Text
React API

API Reference for the Cheetah React SDK (npmjs).

useCheetah()

function useCheetah(): {
  result: {
    transcript: string;
    words?: CheetahWord[];
    isComplete?: boolean;
  } | null;
  isLoaded: boolean;
  isListening: boolean;
  error: Error | null;
  init: (
    accessKey: string,
    model: CheetahModel,
    options?: CheetahOptions
  ) => Promise<void>;
  start: () => Promise<void>;
  startAnnotated: () => Promise<void>;
  stop: () => Promise<void>;
  release: () => Promise<void>;
}

React Hook for Cheetah speech-to-text engine.

Returns

• result Object | null : Any newly-transcribed speech and a flag indicating if a transcription is complete.
• result.transcript string : A string value representing newly-transcribed speech. If none is available, an empty string is returned.
• result.words CheetahWord[] | undefined : A list of transcribed words with their associated metadata. Will be undefined, unless processing with annotations.
• result.isComplete boolean | undefined : A flag representing whether the current result represents an endpoint (a chunk of audio after an utterance without any speech in it).
• isLoaded boolean : A state indicating whether the engine is initialized successfully.
• isListening boolean : A state indicating whether the webVoiceProcessor is passing audio to the engine.
• error Error | null : An Error object containing details about the error.
• init (...) => Promise<void> : A method to initialize Cheetah given the arguments. See init.
• start (...) => Promise<void> : A method to start recording and processing audio. See start.
• startAnnotated (...) => Promise<void> : A method to start recording and processing audio with annotations. See startAnnotated.
• stop () => Promise<void> : A method to stop recording and processing audio. See stop.
• release () => Promise<void> : Releases resources acquired by useCheetah. See release.

init()

async function init(
  accessKey: string,
  model: CheetahModel,
  options: CheetahOptions = {},
): Promise<void>

Initializes useCheetah hook. Sets isLoaded to true.

Parameters

• accessKey string : AccessKey obtained from Picovoice Console.
• model CheetahModel : Object containing Cheetah model.
• options CheetahOptions : Optional configuration arguments.

start()

async function start(): Promise<void>

Start recording and processing audio. If started, isListening is set to true.

startAnnotated()

async function startAnnotated(): Promise<void>

Start recording and processing audio with annotations. If started, isListening is set to true.

stop()

async function stop(): Promise<void>

Stop recording and processing audio. Sets isListening to false. Will resolve its promise after the flushed frame is returned.

release()

function release(): Promise<void>

Releases resources acquired by the useCheetah.