mugib_voice 0.2.1

Official Flutter SDK for Mugib voice agents. One client for VAPI and Google Gemini Live — pass apiKey and agentId only.

mugib_voice

Official Flutter SDK for Mugib real-time voice agents.

Talk to your AI agent by voice with a single client. The SDK includes VAPI and Gemini Live support — you only pass apiKey + agentId. The SDK calls /token and picks the right provider automatically (same flow as voice-widget.js).

---

Install

Add to your app's pubspec.yaml:

dependencies:
  mugib_voice: ^0.2.0

Then:

flutter pub get

Monorepo / local dev: use path: ../sdk/flutter/mugib_voice instead of the version constraint above.

Requirements: Flutter 3.24+ / Dart 3.5+

Bundled dependencies:

• vapi — VAPI WebRTC calls
• web_socket_channel — Gemini Live stream
• record — microphone PCM capture
• flutter_pcm_sound — Gemini PCM playback
• http — agent list + token API

---

Platform setup

iOS — ios/Runner/Info.plist

<key>NSMicrophoneUsageDescription</key>
<string>We need the microphone for voice conversations.</string>

Set the minimum iOS version to 12.0+ in ios/Podfile:

platform :ios, '12.0'

Android — android/app/src/main/AndroidManifest.xml

<uses-permission android:name="android.permission.RECORD_AUDIO" />
<uses-permission android:name="android.permission.INTERNET" />

Set minSdkVersion 23 (or higher) in android/app/build.gradle.

---

Quick start

import 'package:mugib_voice/mugib_voice.dart';

// Once at app startup (initializes VAPI native layers)
await MugibVoice.initialize();

// List agents to show a picker (includes provider per agent)
final agents = await MugibVoice.listAgents(apiKey: 'YOUR_PROJECT_API_KEY');
for (final a in agents) {
  print('${a.id} ${a.name} provider=${a.provider}');
}

final call = MugibVoiceClient(
  apiKey: 'YOUR_PROJECT_API_KEY',
  agentId: 34,
  // baseUrl: 'https://api.mugib.com',
);

call.state.listen((s) => print('state: $s'));
call.transcripts.listen((t) => print('${t.role}: ${t.text}'));
call.errors.listen((e) => print('error: $e'));

await call.start(); // fetches /token → VAPI or Gemini Live

call.mute();
call.unmute();

await call.end();
await call.dispose();

---

API

MugibVoice

Method	Description
Future<void> initialize()	Initialize VAPI native/WebRTC (call once in main)
Future<List<MugibAgentSummary>> listAgents({apiKey, baseUrl})	GET /api/v1/voice-agents — includes provider

MugibVoiceClient(...)

Param	Type	Default	Description
apiKey	String	—	Project API key
agentId	int	—	Voice agent id
baseUrl	String	https://api.mugib.com	API base URL

Method	Description
Future<void> start()	Fetches /token, connects via VAPI or Gemini Live
Future<void> end()	Ends the call
void mute() / unmute() / bool toggleMute()	Microphone control
Future<void> dispose()	Releases all resources
MugibVoiceToken? get token	Last resolved token (provider, agent name, …)

Streams

Stream	Emits
state	MugibCallState
transcripts	MugibTranscript { role, text }
errors	String

MugibAgentSummary (from list endpoint)

Field	Description
id, name, language, dialect, voiceGender	Display / picker
provider	"vapi" or "gemini_live"
isActive	Always true in list (inactive agents are filtered server-side)

---

How it works

App (apiKey + agentId)
  → GET /api/v1/voice-agents/{id}/token
  → provider = vapi        → VapiClient.start(assistantId)
  → provider = gemini_live → WebSocket /stream + PCM mic/playback

Knowledge base, prompts, language, and voice are resolved server-side.

---

Publish / update (maintainers)

cd sdk/flutter/mugib_voice
make login          # once
make release-patch  # bump patch + publish to pub.dev
# or: make update   # publish current version only
make help

Verify (maintainers)

cd sdk/flutter/mugib_voice
make check

---

License

MIT — see LICENSE.