Skip to main content

Audio transcription and generation

Transcribe spoken language from audio files and create spoken audio using AI-generated speech​

Use cases​

  • Automatically trigger transcription when new audio files are uploaded.
  • Save and summarize transcriptions using a Squid AI Agent.
  • Create an audible persona for your AI agent.

Run transcription​

To transcribe a file, use the Squid Client SDK.

Note

The Squid AI Audio client requires admin access to your Squid resources. It should only be used in a secure environment where you can safely provide your Squid API key, such as the Squid backend.

To transcribe a file, use Squid AI Audio's transcribe() method, passing the audio file data:

Backend code
const fileName = 'myAudioFile.mp3';

const audioBlobAndFilename = {
  audioBlob, // provide your audio as a Blob
  fileName,
};
const transcription = await squid.ai().audio().transcribe(audioBlobAndFilename, {
  modelName: 'whisper-1',
});

The transcribe() method also accepts an optional options parameter for customizing your transcription. More information can be found in the reference documentation.

Create audio files​

To create AI-generated audio files, use Squid AI Audio's createSpeech() method. This method takes an input string and an options parameter, which is used to customize the audio file. The method returns a promise that resolves to the generated audio file.

Backend code
const audioFile = await squid.ai().audio().createSpeech(
  'Say hello to all users like a pirate would say hello.',
  { modelName: 'tts-1' },
);