Class DeepgramTextToSpeechProvider

Text-to-speech provider backed by the Deepgram Aura /v1/speak REST API.

API Contract

• Endpoint: POST {baseUrl}/speak?model={voice}&encoding={enc}
• Authentication: Authorization: Token <apiKey>
• Request body: { text }
• Response: Raw audio bytes (audio/mpeg for mp3)

Aura caps a request at 2000 characters, so longer text is chunked at sentence boundaries and the audio buffers are concatenated.

Example

const provider = new DeepgramTextToSpeechProvider({
  apiKey: process.env.DEEPGRAM_API_KEY!,
  voice: 'aura-2-arcas-en',
});
const result = await provider.synthesize('Hello there!');
Copy