HaishinKit for iOS, macOS, tvOS, visionOS and Android.

• Camera and Microphone streaming library via RTMP and SRT for iOS, macOS, tvOS and visionOS.
• README.md contains unreleased content, which can be tested on the main branch.
• API Documentation

💖 Sponsors

Sponsorship

💬 Communication

• If you need help with making LiveStreaming requests using HaishinKit, use a GitHub Discussions with Q&A.
• If you'd like to discuss a feature request, use a GitHub Discussions with Idea
• If you met a HaishinKit's bug🐛, use a GitHub Issue with Bug report template
  • The trace level log is very useful. Please set LBLogger.with(HaishinKitIdentifier).level = .trace.
  • If you don't use an issue template. I will immediately close the your issue without a comment.
• If you want to contribute, submit a pull request with a pr template.
• If you want to support e-mail based communication without GitHub.
  • Consulting fee is $50/1 incident. I'm able to response a few days.
• Discord chatroom.
• 日本語が分かる方は、日本語でのコミニケーションをお願いします！

🌏 Related projects

🎨 Features

RTMP

• [x] Authentication
• [x] Publish and Recording
• [x] Playback (Beta)
• [x] Adaptive bitrate streaming
• [ ] Action Message Format
  • [x] AMF0
  • [ ] AMF3
• [x] SharedObject
• [x] RTMPS
  • [x] Native (RTMP over SSL/TLS)
  • [x] Tunneled (RTMPT over SSL/TLS) (Technical Preview)
• [x] RTMPT (Technical Preview)
• [x] ReplayKit Live as a Broadcast Upload Extension
• [x] Enhanced RTMP

SRT(beta)

• [x] Publish and Recording (H264/HEVC/AAC)
• [x] Playback(beta)
• [ ] mode
  • [x] caller
  • [ ] listener
  • [ ] rendezvous

Offscreen Rendering.

Through off-screen rendering capabilities, it is possible to display any text or bitmap on a video during broadcasting or viewing. This allows for various applications such as watermarking and time display.

Rendering

Others

• [x] tvOS 17.0 for AVCaptureSession.
• [x] Support multitasking camera access.
• [x] Support "Allow app extension API only" option

🐾 Examples

Examples project are available for iOS with UIKit, iOS with SwiftUI, macOS and tvOS. Example macOS requires Apple Silicon mac.

• [x] Camera and microphone publish.
• [x] Playback

  git clone https://github.com/shogo4405/HaishinKit.swift.git
  cd HaishinKit.swift
  carthage bootstrap --platform iOS,macOS,tvOS --use-xcframeworks
  open HaishinKit.xcodeproj

🌏 Requirements

Development

OS

Cocoa Keys

Please contains Info.plist.

iOS 10.0+

• NSMicrophoneUsageDescription
• NSCameraUsageDescription

macOS 10.14+

• NSMicrophoneUsageDescription
• NSCameraUsageDescription

tvOS 17.0+

• NSMicrophoneUsageDescription
• NSCameraUsageDescription

🔧 Installation

source 'https://github.com/CocoaPods/Specs.git'
use_frameworks!

def import_pods
  pod 'HaishinKit', '~> 1.9.1
end

target 'Your Target'  do
  platform :ios, '13.0'
  import_pods
end

source 'https://github.com/CocoaPods/Specs.git'
use_frameworks!

def import_pods
  pod 'SRTHaishinKit', '~> 1.9.1
end

target 'Your Target'  do
  platform :ios, '13.0'
  import_pods
end

🔧 Prerequisites

Make sure you setup and activate your AVAudioSession iOS.

import AVFoundation
let session = AVAudioSession.sharedInstance()
do {
    try session.setCategory(.playAndRecord, mode: .default, options: [.defaultToSpeaker, .allowBluetooth])
    try session.setActive(true)
} catch {
    print(error)
}

📓 RTMP Usage

Ingest

let connection = RTMPConnection()
let stream = RTMPStream(connection: connection)

stream.attachAudio(AVCaptureDevice.default(for: .audio)) { _, error in
  if let error {
    logger.warn(error)
  }
}

stream.attachCamera(AVCaptureDevice.default(.builtInWideAngleCamera, for: .video, position: .back), track: 0) { _, error in
  if let error {
    logger.warn(error)
  }
}

let hkView = MTHKView(frame: view.bounds)
hkView.videoGravity = AVLayerVideoGravity.resizeAspectFill
hkView.attachStream(stream)

// add ViewController#view
view.addSubview(hkView)

connection.connect("rtmp://localhost/appName/instanceName")
stream.publish("streamName")

Playback

let connection = RTMPConnection()
let stream = RTMPStream(connection: connection)

let hkView = MTHKView(frame: view.bounds)
hkView.videoGravity = AVLayerVideoGravity.resizeAspectFill
hkView.attachStream(stream)

// add ViewController#view
view.addSubview(hkView)

connection.connect("rtmp://localhost/appName/instanceName")
stream.play("streamName")

Authentication

var connection = RTMPConnection()
connection.connect("rtmp://username:password@localhost/appName/instanceName")

📓 SRT Usage

Ingest

let connection = SRTConnection()
let stream = SRTStream(connection: connection)
stream.attachAudio(AVCaptureDevice.default(for: .audio)) { error in
    // print(error)
}
stream.attachCamera(AVCaptureDevice.default(.builtInWideAngleCamera, for: .video, position: .back), track: 0) { _, error in
  if let error {
    logger.warn(error)
  }
}

let hkView = HKView(frame: view.bounds)
hkView.videoGravity = AVLayerVideoGravity.resizeAspectFill
hkView.attachStream(rtmpStream)

// add ViewController#view
view.addSubview(hkView)

connection.connect("srt://host:port?option=foo")
stream.publish()

Playback

let connection = SRTConnection()
let stream = SRTStream(connection: connection)

let hkView = MTHKView(frame: view.bounds)
hkView.videoGravity = AVLayerVideoGravity.resizeAspectFill
hkView.attachStream(rtmpStream)

// add ViewController#view
view.addSubview(hkView)

connection.connect("srt://host:port?option=foo")
stream.play()

📓 Settings

📹 AVCaptureSession

stream.frameRate = 30
stream.sessionPreset = AVCaptureSession.Preset.medium

// Do not call beginConfiguration() and commitConfiguration() internally within the scope of the method, as they are called internally.
stream.configuration { session in
  session.automaticallyConfiguresApplicationAudioSession = true
}

🔊 Audio

Capture

Specifies the capture capture settings.

let front = AVCaptureDevice.default(for: .audio)
stream.attachAudio(front, track: 0) { audioUnit, error in
}

AudioMixerSettings

If you want to mix multiple audio tracks, please enable the feature flag.

stream.isMultiTrackAudioMixingEnabled = true

When you specify the sampling rate, it will perform resampling. Additionally, in the case of multiple channels, downsampling can be applied.

// Setting the value to 0 will be the same as the value specified in mainTrack.
stream.audioMixerSettings = IOAudioMixerSettings(
  sampleRate: Float64 = 44100,
  channels: UInt32 = 0,
)

stream.audioMixerSettings.isMuted = false
stream.audioMixerSettings.mainTrack = 0
stream.audioMixerSettings.tracks = [
  0: .init(
    isMuted: Bool = false,
    downmix: Bool = true,
    channelMap: [Int]? = nil
  )
]

AudioCodecSettings

/// Specifies the bitRate of audio output.
stream.audioSettings.bitrate = 64 * 1000
/// Specifies the mixes the channels or not. Currently, it supports input sources with 4, 5, 6, and 8 channels.
stream.audioSettings.downmix = true
/// Specifies the map of the output to input channels.
 stream.audioSettings.channelMap: [Int]? = nil

🎥 Video

Capture

Specifies the video capture settings.

let front = AVCaptureDevice.default(.builtInWideAngleCamera, for: .video, position: .front)
stream.attachCamera(front, track: 0) { videoUnit, error in
  videoUnit?.isVideoMirrored = true
  videoUnit?.preferredVideoStabilizationMode = .standard
  videoUnit?.colorFormat = kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange
}

VideoMixerSettings

/// Specifies the image rendering mode.
stream.videoMixerSettings.mode = .passthrough or .offscreen
/// Specifies the muted indicies whether freeze video signal or not.
stream.videoMixerSettings.isMuted = false
/// Specifies the main track number.
stream.videoMixerSettings.mainTrack = 0

VideoCodecSettings

stream.videoSettings = .init(
  videoSize: .init(width: 854, height: 480),
  profileLevel: kVTProfileLevel_H264_Baseline_3_1 as String,
  bitRate: 640 * 1000,
  maxKeyFrameIntervalDuration: 2,
  scalingMode: .trim,
  bitRateMode: .average,
  allowFrameReordering: nil,
  isHardwareEncoderEnabled: true
)

⏺️ Recording

Internally, I am now handling data with more than 3 channels. If you encounter audio issues with IOStreamRecorder, it is recommended to set it back to a maximum of 2 channels when saving locally.

let channels = max(stream.audioInputFormats[0].channels ?? 1, 2)
stream.audioMixerSettings = .init(sampleRate: 0, channels: channels)

// Specifies the recording settings. 0" means the same of input.
var recorder = IOStreamRecorder()
stream.addObserver(recorder)

recorder.settings = [
  AVMediaType.audio: [
    AVFormatIDKey: Int(kAudioFormatMPEG4AAC),
    AVSampleRateKey: 0,
    AVNumberOfChannelsKey: 0,
    // AVEncoderBitRateKey: 128000,
  ],
  AVMediaType.video: [
    AVVideoCodecKey: AVVideoCodecH264,
    AVVideoHeightKey: 0,
    AVVideoWidthKey: 0,
    /*
    AVVideoCompressionPropertiesKey: [
      AVVideoMaxKeyFrameIntervalDurationKey: 2,
      AVVideoProfileLevelKey: AVVideoProfileLevelH264Baseline30,
      AVVideoAverageBitRateKey: 512000
    ]
    */
  ]
]

recorder.startRunning()
// recorder.stopRunning()

📜 License

BSD-3-Clause