Merge pull request 'Fix listening practice crash on Start Speaking' (#16) from fix/listening-crash into main

Wrap startRecording in do/catch so audio setup failures don't crash.
Validate recording format has channels before installTap. Use
DispatchQueue.main.async instead of Task{@MainActor} in recognition
callback to avoid dispatch queue assertions.

Closes #13

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Conjuga/Conjuga/Services/PronunciationService.swift

@@ -50,45 +50,64 @@ final class PronunciationService {
         recognizer = SFSpeechRecognizer(locale: Locale(identifier: "es-ES"))
     }
 
-    func startRecording() throws {
-        guard isAuthorized else { return }
+    func startRecording() {
+        guard isAuthorized else {
+            print("[PronunciationService] not authorized")
+            return
+        }
         resolveRecognizerIfNeeded()
-        guard let recognizer, recognizer.isAvailable else { return }
+        guard let recognizer, recognizer.isAvailable else {
+            print("[PronunciationService] recognizer unavailable")
+            return
+        }
 
         stopRecording()
 
-        let audioSession = AVAudioSession.sharedInstance()
-        try audioSession.setCategory(.playAndRecord, mode: .measurement, options: [.duckOthers, .defaultToSpeaker])
-        try audioSession.setActive(true, options: .notifyOthersOnDeactivation)
+        do {
+            let audioSession = AVAudioSession.sharedInstance()
+            try audioSession.setCategory(.playAndRecord, mode: .measurement, options: [.duckOthers, .defaultToSpeaker])
+            try audioSession.setActive(true, options: .notifyOthersOnDeactivation)
 
-        audioEngine = AVAudioEngine()
-        request = SFSpeechAudioBufferRecognitionRequest()
+            audioEngine = AVAudioEngine()
+            request = SFSpeechAudioBufferRecognitionRequest()
 
-        guard let audioEngine, let request else { return }
-        request.shouldReportPartialResults = true
+            guard let audioEngine, let request else { return }
+            request.shouldReportPartialResults = true
 
-        let inputNode = audioEngine.inputNode
-        let recordingFormat = inputNode.outputFormat(forBus: 0)
+            let inputNode = audioEngine.inputNode
+            let recordingFormat = inputNode.outputFormat(forBus: 0)
 
-        inputNode.installTap(onBus: 0, bufferSize: 1024, format: recordingFormat) { buffer, _ in
-            request.append(buffer)
-        }
+            // Validate format — 0 channels crashes installTap
+            guard recordingFormat.channelCount > 0 else {
+                print("[PronunciationService] invalid recording format (0 channels)")
+                self.audioEngine = nil
+                self.request = nil
+                return
+            }
 
-        audioEngine.prepare()
-        try audioEngine.start()
+            inputNode.installTap(onBus: 0, bufferSize: 1024, format: recordingFormat) { buffer, _ in
+                request.append(buffer)
+            }
 
-        transcript = ""
-        isRecording = true
+            audioEngine.prepare()
+            try audioEngine.start()
 
-        task = recognizer.recognitionTask(with: request) { [weak self] result, error in
-            Task { @MainActor in
-                if let result {
-                    self?.transcript = result.bestTranscription.formattedString
-                }
-                if error != nil || (result?.isFinal == true) {
-                    self?.stopRecording()
+            transcript = ""
+            isRecording = true
+
+            task = recognizer.recognitionTask(with: request) { [weak self] result, error in
+                DispatchQueue.main.async {
+                    if let result {
+                        self?.transcript = result.bestTranscription.formattedString
+                    }
+                    if error != nil || (result?.isFinal == true) {
+                        self?.stopRecording()
+                    }
                 }
             }
+        } catch {
+            print("[PronunciationService] startRecording failed: \(error)")
+            stopRecording()
         }
     }
 

Conjuga/Conjuga/Views/Practice/ListeningView.swift

@@ -166,7 +166,7 @@ struct ListeningView: View {
                     if result.score >= 0.7 { correctCount += 1 }
                     withAnimation { isRevealed = true }
                 } else {
-                    try? pronunciation.startRecording()
+                    pronunciation.startRecording()
                 }
             } label: {
                 Label(pronunciation.isRecording ? "Stop" : "Start Speaking", systemImage: pronunciation.isRecording ? "stop.circle.fill" : "mic.circle.fill")