[Enhancement]Improved audio session management (#906)

Author: ipavlidakis

CHANGELOG.md

@@ -4,7 +4,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
 # Upcoming
 
-### 🔄 Changed
+### 🐞 Fixed
+- AudioSession management issues that were causing audio not being recorded during calls. [#906](https://github.com/GetStream/stream-video-swift/pull/906)
 
 # [1.29.1](https://github.com/GetStream/stream-video-swift/releases/tag/1.29.1)
 _July 25, 2025_

DemoApp/Sources/ViewModifiers/MoreControls/DemoMoreControlsViewModifier.swift

@@ -52,6 +52,17 @@ struct DemoMoreControlsViewModifier: ViewModifier {
                                 )
                             }
 
+                            DemoMoreControlListButtonView(
+                                action: { viewModel.toggleAudioOutput() },
+                                label: viewModel.callSettings.audioOutputOn ? "Disable audio output" : "Enable audio output"
+                            ) {
+                                Image(
+                                    systemName: viewModel.callSettings.audioOutputOn
+                                        ? "speaker.fill"
+                                        : "speaker.slash"
+                                )
+                            }
+
                             DemoTranscriptionAndClosedCaptionsButtonView(viewModel: viewModel)
 
                             DemoMoreThermalStateButtonView()

Sources/StreamVideo/Call.swift

@@ -147,6 +147,24 @@ public class Call: @unchecked Sendable, WSEventsSubscriber {
         notify: Bool = false,
         callSettings: CallSettings? = nil
     ) async throws -> JoinCallResponse {
+        /// Determines the source from which the join action was initiated.
+        ///
+        /// This block checks if the `joinSource` has already been set in the current
+        /// call state. If not, it assigns `.inApp` as the default join source,
+        /// indicating the call was joined from within the app UI. The resolved
+        /// `JoinSource` value is then used to record how the call was joined,
+        /// enabling analytics and behavioral branching based on entry point.
+        let joinSource = await {
+            if let joinSource = await state.joinSource {
+                return joinSource
+            } else {
+                return await Task { @MainActor in
+                    state.joinSource = .inApp
+                    return .inApp
+                }.value
+            }
+        }()
+
         let result: Any? = stateMachine.withLock { currentStage, transitionHandler in
             if
                 currentStage.id == .joined,
@@ -194,6 +212,7 @@ public class Call: @unchecked Sendable, WSEventsSubscriber {
                                 options: options,
                                 ring: ring,
                                 notify: notify,
+                                source: joinSource,
                                 deliverySubject: deliverySubject
                             )
                         )
@@ -1371,8 +1390,8 @@ public class Call: @unchecked Sendable, WSEventsSubscriber {
     /// - Parameter policy: A conforming `AudioSessionPolicy` that defines
     ///   the audio session configuration to be applied.
     /// - Throws: An error if the update fails.
-    public func updateAudioSessionPolicy(_ policy: AudioSessionPolicy) async throws {
-        try await callController.updateAudioSessionPolicy(policy)
+    public func updateAudioSessionPolicy(_ policy: AudioSessionPolicy) async {
+        await callController.updateAudioSessionPolicy(policy)
     }
 
     /// Adds a proximity policy to manage device proximity behavior during the call.
@@ -1473,22 +1492,6 @@ public class Call: @unchecked Sendable, WSEventsSubscriber {
         )
     }
 
-    // MARK: - CallKit
-
-    /// Notifies the `Call` instance that CallKit has activated the system audio
-    /// session.
-    ///
-    /// This method should be called when the system activates the `AVAudioSession`
-    /// as a result of an incoming or outgoing CallKit-managed call. It allows the
-    /// call to update the provided CallKit AVAudioSession based on the internal CallSettings.
-    ///
-    /// - Parameter audioSession: The active `AVAudioSession` instance provided by
-    ///   CallKit.
-    /// - Throws: An error if the call controller fails to handle the activation.
-    internal func callKitActivated(_ audioSession: AVAudioSessionProtocol) throws {
-        try callController.callKitActivated(audioSession)
-    }
-
     internal func didPerform(_ action: WebRTCTrace.CallKitAction) {
         Task(disposableBag: disposableBag) { [weak callController] in
             await callController?.didPerform(action)

Sources/StreamVideo/CallKit/CallKitService.swift

@@ -15,6 +15,7 @@ open class CallKitService: NSObject, CXProviderDelegate, @unchecked Sendable {
     @Injected(\.callCache) private var callCache
     @Injected(\.uuidFactory) private var uuidFactory
     @Injected(\.currentDevice) private var currentDevice
+    @Injected(\.audioStore) private var audioStore
     private let disposableBag = DisposableBag()
 
     /// Represents a call that is being managed by the service.
@@ -95,6 +96,13 @@ open class CallKitService: NSObject, CXProviderDelegate, @unchecked Sendable {
     private var callEndedNotificationCancellable: AnyCancellable?
     private var ringingTimerCancellable: AnyCancellable?
 
+    /// A reducer responsible for handling audio session changes triggered by CallKit.
+    ///
+    /// The `callKitAudioReducer` manages updates to the audio session state in
+    /// response to CallKit events, ensuring proper activation and deactivation
+    /// of the audio system when calls are handled through CallKit.
+    private lazy var callKitAudioReducer = CallKitAudioSessionReducer(store: audioStore)
+
     /// Initializes the `CallKitService` instance.
     override public init() {
         super.init()
@@ -164,6 +172,7 @@ open class CallKitService: NSObject, CXProviderDelegate, @unchecked Sendable {
                 return
             }
             do {
+
                 if streamVideo.state.connection != .connected {
                     let result = await Task(disposableBag: disposableBag) { [weak self] in
                         try await self?.streamVideo?.connect()
@@ -392,17 +401,11 @@ open class CallKitService: NSObject, CXProviderDelegate, @unchecked Sendable {
             subsystems: .callKit
         )
 
-        if
-            let active,
-            let call = callEntry(for: active)?.call {
-            call.didPerform(.didActivateAudioSession)
-
-            do {
-                try call.callKitActivated(audioSession)
-            } catch {
-                log.error(error, subsystems: .callKit)
-            }
-        }
+        /// Activates the audio session for CallKit. This line notifies the audio store
+        /// to activate the provided AVAudioSession, ensuring that the app's audio
+        /// routing and configuration are correctly handled when CallKit takes control
+        /// of the audio session during a call.
+        audioStore.dispatch(.callKit(.activate(audioSession)))
     }
 
     public func provider(
@@ -421,11 +424,11 @@ open class CallKitService: NSObject, CXProviderDelegate, @unchecked Sendable {
             """,
             subsystems: .callKit
         )
-        if
-            let active,
-            let call = callEntry(for: active)?.call {
-            call.didPerform(.didDeactivateAudioSession)
-        }
+
+        /// Notifies the audio store to deactivate the provided AVAudioSession.
+        /// This ensures that when CallKit relinquishes control of the audio session,
+        /// the app's audio routing and configuration are updated appropriately.
+        audioStore.dispatch(.callKit(.deactivate(audioSession)))
     }
 
     open func provider(
@@ -460,6 +463,10 @@ open class CallKitService: NSObject, CXProviderDelegate, @unchecked Sendable {
             }
 
             do {
+                /// Sets the join source to `.callKit` to indicate that the call was
+                /// joined via CallKit. This helps with audioSession management.
+                callToJoinEntry.call.state.joinSource = .callKit
+
                 try await callToJoinEntry.call.join(callSettings: callSettings)
                 action.fulfill()
             } catch {
@@ -640,9 +647,16 @@ open class CallKitService: NSObject, CXProviderDelegate, @unchecked Sendable {
     /// A method that's being called every time the StreamVideo instance is getting updated.
     /// - Parameter streamVideo: The new StreamVideo instance (nil if none)
     open func didUpdate(_ streamVideo: StreamVideo?) {
+        if streamVideo != nil {
+            audioStore.add(callKitAudioReducer)
+        } else {
+            audioStore.remove(callKitAudioReducer)
+        }
+
         guard currentDevice.deviceType != .simulator else {
             return
         }
+
         subscribeToCallEvents()
     }
 

Sources/StreamVideo/CallState.swift

@@ -155,7 +155,14 @@ public class CallState: ObservableObject {
             }
         }
     }
-    
+
+    /// Describes the source from which the join action was triggered for this call.
+    ///
+    /// Use this property to determine whether the current call was joined from
+    /// the app's UI or via a system-level integration such as CallKit. This can
+    /// help customize logic, analytics, and UI based on how the call was started.
+    var joinSource: JoinSource?
+
     private var localCallSettingsUpdate = false
     private var durationCancellable: AnyCancellable?
     private nonisolated let disposableBag = DisposableBag()

Sources/StreamVideo/CallStateMachine/Stages/Call+JoiningStage.swift

@@ -122,7 +122,8 @@ extension Call.StateMachine.Stage {
                 callSettings: input.callSettings,
                 options: input.options,
                 ring: input.ring,
-                notify: input.notify
+                notify: input.notify,
+                source: input.source
             )
 
             if let callSettings = input.callSettings {

Sources/StreamVideo/CallStateMachine/Stages/Call+Stage.swift

@@ -30,6 +30,7 @@ extension Call.StateMachine {
                 var options: CreateCallOptions?
                 var ring: Bool
                 var notify: Bool
+                var source: JoinSource
                 var deliverySubject: PassthroughSubject<JoinCallResponse, Error>
 
                 var currentNumberOfRetries = 0

Sources/StreamVideo/Controllers/CallController.swift

@@ -105,24 +105,29 @@ class CallController: @unchecked Sendable {
             .sinkTask(storeIn: disposableBag) { [weak self] in await self?.didFetch($0) }
     }
 
-    /// Joins a call with the provided information.
+    /// Joins a call with the provided information and join source.
+    ///
     /// - Parameters:
-    ///  - callType: the type of the call
-    ///  - callId: the id of the call
-    ///  - callSettings: the current call settings
-    ///  - videoOptions: configuration options about the video
-    ///  - options: create call options
-    ///  - migratingFrom: if SFU migration is being performed
-    ///  - ring: whether ringing events should be handled
-    ///  - notify: whether uses should be notified about the call
-    /// - Returns: a newly created `Call`.
+    ///   - callType: The type of the call.
+    ///   - callId: The id of the call.
+    ///   - callSettings: The current call settings.
+    ///   - videoOptions: Configuration options about the video.
+    ///   - options: Create call options.
+    ///   - migratingFrom: If SFU migration is being performed.
+    ///   - ring: Whether ringing events should be handled.
+    ///   - notify: Whether users should be notified about the call.
+    ///   - source: Describes the source from which the join action was triggered.
+    ///            Use this to indicate if the call was joined from in-app UI or
+    ///            via CallKit.
+    /// - Returns: A newly created `JoinCallResponse`.
     @discardableResult
     func joinCall(
         create: Bool = true,
         callSettings: CallSettings?,
         options: CreateCallOptions? = nil,
         ring: Bool = false,
-        notify: Bool = false
+        notify: Bool = false,
+        source: JoinSource
     ) async throws -> JoinCallResponse {
         joinCallResponseSubject = .init(nil)
 
@@ -131,7 +136,8 @@ class CallController: @unchecked Sendable {
             callSettings: callSettings,
             options: options,
             ring: ring,
-            notify: notify
+            notify: notify,
+            source: source
         )
 
         guard
@@ -479,8 +485,8 @@ class CallController: @unchecked Sendable {
     ///
     /// - Parameter policy: The audio session policy to apply
     /// - Throws: An error if the policy update fails
-    func updateAudioSessionPolicy(_ policy: AudioSessionPolicy) async throws {
-        try await webRTCCoordinator.updateAudioSessionPolicy(policy)
+    func updateAudioSessionPolicy(_ policy: AudioSessionPolicy) async {
+        await webRTCCoordinator.updateAudioSessionPolicy(policy)
     }
 
     /// Sets up observation of WebRTC state changes.
@@ -501,10 +507,6 @@ class CallController: @unchecked Sendable {
             .sink { [weak self] in self?.webRTCClientDidUpdateStage($0) }
     }
 
-    internal func callKitActivated(_ audioSession: AVAudioSessionProtocol) throws {
-        try webRTCCoordinator.callKitActivated(audioSession)
-    }
-
     // MARK: - Client Capabilities
 
     func enableClientCapabilities(_ capabilities: Set<ClientCapability>) async {

Sources/StreamVideo/Models/JoinSource.swift

@@ -0,0 +1,19 @@
+//
+// Copyright © 2025 Stream.io Inc. All rights reserved.
+//
+
+import Foundation
+
+/// An enumeration that describes the source from which a call was joined.
+///
+/// Use `JoinSource` to indicate whether the join action originated from within
+/// the app's own UI or through a system-level interface such as CallKit.
+/// This helps distinguish the user's entry point and can be used to customize
+/// behavior or analytics based on how the call was initiated.
+enum JoinSource {
+    /// Indicates that the call was joined from within the app's UI.
+    case inApp
+
+    /// Indicates that the call was joined via CallKit integration.
+    case callKit
+}

Sources/StreamVideo/StreamVideo.swift

@@ -16,6 +16,7 @@ public class StreamVideo: ObservableObject, @unchecked Sendable {
 
     @Injected(\.callCache) private var callCache
     @Injected(\.screenProperties) private var screenProperties
+    @Injected(\.audioStore) private var audioStore
 
     private enum DisposableKey: String { case ringEventReceived }