SDKs

Hakim AI ships two official SDKs, both wrapping the same public REST/WebSocket API documented in the reference.

TypeScript / JavaScript

npm install @tryhakim/voice

import { Hakim } from '@tryhakim/voice';

const hakim = new Hakim({ apiKey: process.env.HAKIM_API_KEY! });

const speech = await hakim.audio.speech.create({
  model: 'hakim-fast-v1',
  input: 'مرحبا، أنا حكيم.',
  voice: 'your-voice-id',
  response_format: 'mp3',
});

Streaming synthesis (low-latency start):

const { stream } = await hakim.audio.speech.stream({
  model: 'hakim-fast-v1',
  input: 'مرحبا',
  voice: 'your-voice-id',
});
for await (const chunk of stream) {
  process.stdout.write(chunk);
}

Source and full API surface: github.com/tryHakimAI/hakim-node.

Python

pip install hakim

from hakim import Hakim

client = Hakim(api_key=os.environ["HAKIM_API_KEY"])
audio = client.audio.speech.create(
    model="hakim-fast-v1",
    input="مرحبا، أنا حكيم.",
    voice="your-voice-id",
)

Ships both a sync (Hakim) and async (AsyncHakim) client. Source and full API surface: github.com/tryHakimAI/hakim-python.

Voice agent integrations

Building a realtime voice agent instead of calling the API directly? Drop-in plugins are available for two popular open-source frameworks:

Both talk to Hakim's public REST/WebSocket API directly, so they work anywhere the underlying framework runs.

Both SDKs respect the Authorization: Bearer header and retry on 429 according to the Retry-After response.