SDKsuni-app / uni-app xEnterprise

Transcribe audio

Convert audio recordings into transcribed text using the speech-to-text service.

Audio transcription is a Commercial Edition capability. Invoke speechToText() with the audio filename and the audio binary encoded as a pure Base64 string. Do not pass temporary filesystem paths, ArrayBuffer objects, or Data URLs containing data:*;base64, headers.

Parameters

Parameter	Type	Required	Description
`filename`	`string`	Yes	Audio filename with an extension matching the audio format (for example, `'voice.m4a'`).
`data`	`string`	Yes	Raw Base64-encoded audio payload string without data URI prefixes.

Verify that your audio adheres to format, sample rate, and size limits beforehand by invoking Check speech-to-text availability. Read recording files immediately before the OS sweeps temporary directories, or copy them to uni.env.USER_DATA_PATH.

Code example

import { speechToText } from '@/uni_modules/unix-openim-sdk'

const result = await speechToText({
  filename: 'voice.m4a',
  data: audioBase64,
})

const transcript = result?.text ?? ''
showTranscript(transcript)

Return value

The Promise resolves with an OpenIMSpeechToTextResult | null object. The transcribed text string is located in text.

State changes after the call

Transcription output is returned solely as the response to this call; it does not alter the original voice message or emit message events. To cache this transcription text locally on the current device, merge it into the message's localEx via Set message local extension.

Was this page helpful?