feat(audio): replace record package with mic_stream_recorder

2025-11-05 00:09:35 +05:30
parent 24c8c27d51
commit a3b5c4f5b7
3 changed files with 65 additions and 170 deletions
--- a/lib/features/chat/services/voice_input_service.dart
+++ b/lib/features/chat/services/voice_input_service.dart
@@ -4,7 +4,7 @@ import 'dart:io' show File, Platform;
 import 'package:flutter/widgets.dart';
 import 'package:flutter_riverpod/flutter_riverpod.dart';
 import 'package:riverpod_annotation/riverpod_annotation.dart';
-import 'package:record/record.dart';
+import 'package:mic_stream_recorder/mic_stream_recorder.dart';
 import 'package:stts/stts.dart';
 import 'package:path/path.dart' as p;
 import 'package:path_provider/path_provider.dart';
@@ -23,7 +23,7 @@ class LocaleName {
 }

 class VoiceInputService {
-  final AudioRecorder _recorder = AudioRecorder();
+  final MicStreamRecorder _recorder = MicStreamRecorder();
  final Stt _speech = Stt();
  final ApiService? _api;
  bool _isInitialized = false;
@@ -31,14 +31,10 @@ class VoiceInputService {
  bool _localSttAvailable = false;
  SttPreference _preference = SttPreference.auto;
  bool _usingServerStt = false;
-  bool _serverRecorderActive = false;
-  String? _serverRecordingPath;
-  String? _serverRecordingMimeType;
  String? _selectedLocaleId;
  List<LocaleName> _locales = const [];
  StreamController<String>? _textStreamController;
  String _currentText = '';
-  // Public stream for UI waveform visualization (emits partial text length as proxy)
  StreamController<int>? _intensityController;
  Stream<int> get intensityStream =>
      _intensityController?.stream ?? const Stream<int>.empty();
@@ -46,12 +42,13 @@ class VoiceInputService {
  Timer? _intensityDecayTimer;
  Timer? _silenceTimer;
  bool _hasDetectedSpeech = false;
+  int _amplitudeCallbackCount = 0;
+  Timer? _amplitudeFallbackTimer;

-  /// Public stream of partial/final transcript strings and special audio tokens.
  Stream<String> get textStream =>
      _textStreamController?.stream ?? const Stream<String>.empty();
  Timer? _autoStopTimer;
-  StreamSubscription<Amplitude>? _ampSub;
+  StreamSubscription<double>? _ampSub;
  StreamSubscription<SttRecognition>? _sttResultSub;
  StreamSubscription<SttState>? _sttStateSub;

@@ -111,10 +108,7 @@ class VoiceInputService {

  Future<bool> checkPermissions() async {
    try {
-      // Prefer stts permission check which will request microphone permission
-      final mic = await _speech.hasPermission();
-      if (mic) return true;
-      return await _recorder.hasPermission();
+      return await _speech.hasPermission();
    } catch (_) {
      return false;
    }
@@ -200,9 +194,6 @@ class VoiceInputService {
    _intensityController = StreamController<int>.broadcast();
    _lastIntensity = 0;
    _usingServerStt = false;
-    _serverRecorderActive = false;
-    _serverRecordingPath = null;
-    _serverRecordingMimeType = null;

    _startIntensityDecayTimer();

@@ -336,6 +327,9 @@ class VoiceInputService {
    _silenceTimer?.cancel();
    _silenceTimer = null;

+    _amplitudeFallbackTimer?.cancel();
+    _amplitudeFallbackTimer = null;
+
    if (_usingServerStt) {
      await _finalizeServerRecording();
    } else {
@@ -356,9 +350,6 @@ class VoiceInputService {
    await _closeControllers();

    _usingServerStt = false;
-    _serverRecorderActive = false;
-    _serverRecordingPath = null;
-    _serverRecordingMimeType = null;
    _hasDetectedSpeech = false;
  }

@@ -417,52 +408,50 @@ class VoiceInputService {
  }

  Future<void> _startServerRecording() async {
-    final (path, mimeType) = await _createRecordingTarget();
-    _serverRecordingPath = path;
-    _serverRecordingMimeType = mimeType;
-
-    final config = RecordConfig(
-      encoder: AudioEncoder.aacLc,
-      sampleRate: 44100,
-      bitRate: 96000,
-      numChannels: 1,
-      noiseSuppress: true,
-    );
-
-    await _recorder.start(config, path: path);
-    _serverRecorderActive = true;
+    final path = await _createRecordingPath();
    _hasDetectedSpeech = false;

+    await _recorder.startRecording(path);
+
    await _ampSub?.cancel();
-    _ampSub = _recorder
-        .onAmplitudeChanged(const Duration(milliseconds: 140))
-        .listen((Amplitude amplitude) {
+    _amplitudeFallbackTimer?.cancel();
+    _amplitudeCallbackCount = 0;
+
+    _ampSub = _recorder.amplitudeStream.listen((amplitude) {
+      _amplitudeCallbackCount++;
      if (!_isListening) return;
-          _lastIntensity = _amplitudeToIntensity(amplitude.current);
+
+      _lastIntensity = _normalizedToIntensity(amplitude);
      try {
        _intensityController?.add(_lastIntensity);
      } catch (_) {}

-          // Detect silence and auto-stop for server-side STT
-          _handleServerAmplitude(amplitude.current);
-        }, onError: (_) {});
+      _handleServerAmplitude(amplitude);
+    });
+
+    _amplitudeFallbackTimer = Timer(const Duration(seconds: 1), () {
+      if (_amplitudeCallbackCount == 0) {
+        _silenceTimer = Timer(const Duration(seconds: 15), () {
+          if (_isListening && _usingServerStt) {
+            unawaited(_stopListening());
+          }
+        });
+      }
+    });
  }

-  void _handleServerAmplitude(double? amplitude) {
+  void _handleServerAmplitude(double amplitude) {
    if (!_usingServerStt || !_isListening) return;

-    // Threshold for detecting speech (in dB)
-    const double speechThreshold = -45.0;
-    final double currentDb = amplitude ?? -100.0;
+    const double speechThreshold = 0.55;
+    if (amplitude.isNaN || amplitude.isInfinite) return;

-    // If we detect speech, mark it and reset silence timer
-    if (currentDb > speechThreshold) {
+    if (amplitude > speechThreshold) {
      _hasDetectedSpeech = true;
      _silenceTimer?.cancel();
      _silenceTimer = null;
    } else if (_hasDetectedSpeech && _silenceTimer == null) {
-      // Start silence timer only after we've detected speech at least once
-      _silenceTimer = Timer(const Duration(seconds: 2), () {
+      _silenceTimer = Timer(const Duration(milliseconds: 800), () {
        if (_isListening && _usingServerStt) {
          unawaited(_stopListening());
        }
@@ -470,53 +459,30 @@ class VoiceInputService {
    }
  }

-  Future<(String, String)> _createRecordingTarget() async {
+  Future<String> _createRecordingPath() async {
    final directory = await getTemporaryDirectory();
    final timestamp = DateTime.now().millisecondsSinceEpoch;
-    const extension = 'm4a';
-    final fileName = 'conduit_voice_$timestamp.$extension';
-    final path = p.join(directory.path, fileName);
-    return (path, 'audio/mp4');
+    final fileName = 'conduit_voice_$timestamp.m4a';
+    return p.join(directory.path, fileName);
  }

  Future<void> _finalizeServerRecording() async {
    final api = _api;
-    if (api == null) {
-      return;
-    }
+    if (api == null) return;

-    String? path;
+    final path = await _recorder.stopRecording();
+    if (path == null || path.isEmpty) return;
+
+    final file = File(path);
    try {
-      if (_serverRecorderActive && await _recorder.isRecording()) {
-        path = await _recorder.stop();
-      } else {
-        path = _serverRecordingPath;
-      }
-    } catch (_) {
-      path = _serverRecordingPath;
-    } finally {
-      _serverRecorderActive = false;
-    }
-
-    final resolvedPath = path;
-    if (resolvedPath == null || resolvedPath.isEmpty) {
-      return;
-    }
-
-    final file = File(resolvedPath);
-    try {
-      if (!await file.exists()) {
-        return;
-      }
+      if (!await file.exists()) return;
      final bytes = await file.readAsBytes();
-      if (bytes.isEmpty) {
-        return;
-      }
+      if (bytes.isEmpty) return;

      final response = await api.transcribeSpeech(
        audioBytes: bytes,
-        fileName: p.basename(resolvedPath),
-        mimeType: _serverRecordingMimeType,
+        fileName: p.basename(path),
+        mimeType: 'audio/mp4',
        language: _languageForServer(),
      );

@@ -641,21 +607,9 @@ class VoiceInputService {
    return null;
  }

-  int _amplitudeToIntensity(double? value) {
-    if (value == null || value.isNaN || value.isInfinite) {
-      return 0;
-    }
-    const minDb = -55.0;
-    const maxDb = 0.0;
-    final double clamped = value.clamp(minDb, maxDb).toDouble();
-    final double normalized = ((clamped - minDb) / (maxDb - minDb)).clamp(
-      0.0,
-      1.0,
-    );
-    final int scaled = (normalized * 10).round();
-    if (scaled <= 0) return 0;
-    if (scaled >= 10) return 10;
-    return scaled;
+  int _normalizedToIntensity(double value) {
+    if (value.isNaN || value.isInfinite) return 0;
+    return (value * 10).round().clamp(0, 10);
  }

  Future<void> _closeControllers() async {
@@ -693,9 +647,6 @@ class VoiceInputService {
    try {
      _speech.dispose().catchError((_) {});
    } catch (_) {}
-    try {
-      _recorder.dispose().catchError((_) {});
-    } catch (_) {}
  }
 }

--- a/pubspec.lock
+++ b/pubspec.lock
@@ -965,6 +965,14 @@ packages:
      url: "https://pub.dev"
    source: hosted
    version: "1.16.0"
+  mic_stream_recorder:
+    dependency: "direct main"
+    description:
+      name: mic_stream_recorder
+      sha256: "73965991ef5cc93d2b0c1e6d590cbd567a853b9ee7b2d52de43a73f185bb0d9c"
+      url: "https://pub.dev"
+    source: hosted
+    version: "1.1.2"
  mime:
    dependency: transitive
    description:
@@ -1165,70 +1173,6 @@ packages:
      url: "https://pub.dev"
    source: hosted
    version: "1.5.0"
-  record:
-    dependency: "direct main"
-    description:
-      name: record
-      sha256: "9dbc6ff3e784612f90a9b001373c45ff76b7a08abd2bd9fdf72c242320c8911c"
-      url: "https://pub.dev"
-    source: hosted
-    version: "6.1.1"
-  record_android:
-    dependency: transitive
-    description:
-      name: record_android
-      sha256: "854627cd78d8d66190377f98477eee06ca96ab7c9f2e662700daf33dbf7e6673"
-      url: "https://pub.dev"
-    source: hosted
-    version: "1.4.2"
-  record_ios:
-    dependency: transitive
-    description:
-      name: record_ios
-      sha256: "13e241ed9cbc220534a40ae6b66222e21288db364d96dd66fb762ebd3cb77c71"
-      url: "https://pub.dev"
-    source: hosted
-    version: "1.1.2"
-  record_linux:
-    dependency: transitive
-    description:
-      name: record_linux
-      sha256: "235b1f1fb84e810f8149cc0c2c731d7d697f8d1c333b32cb820c449bf7bb72d8"
-      url: "https://pub.dev"
-    source: hosted
-    version: "1.2.1"
-  record_macos:
-    dependency: transitive
-    description:
-      name: record_macos
-      sha256: "2849068bb59072f300ad63ed146e543d66afaef8263edba4de4834fc7c8d4d35"
-      url: "https://pub.dev"
-    source: hosted
-    version: "1.1.1"
-  record_platform_interface:
-    dependency: transitive
-    description:
-      name: record_platform_interface
-      sha256: b0065fdf1ec28f5a634d676724d388a77e43ce7646fb049949f58c69f3fcb4ed
-      url: "https://pub.dev"
-    source: hosted
-    version: "1.4.0"
-  record_web:
-    dependency: transitive
-    description:
-      name: record_web
-      sha256: "4f0adf20c9ccafcc02d71111fd91fba1ca7b17a7453902593e5a9b25b74a5c56"
-      url: "https://pub.dev"
-    source: hosted
-    version: "1.2.0"
-  record_windows:
-    dependency: transitive
-    description:
-      name: record_windows
-      sha256: "223258060a1d25c62bae18282c16783f28581ec19401d17e56b5205b9f039d78"
-      url: "https://pub.dev"
-    source: hosted
-    version: "1.0.7"
  riverpod:
    dependency: transitive
    description:
--- a/pubspec.yaml
+++ b/pubspec.yaml
@@ -44,7 +44,7 @@ dependencies:
  flutter_animate: ^4.5.0
  
  # Platform Features
-  record: ^6.1.1
+  mic_stream_recorder: ^1.1.2
  stts: ^1.2.5
  flutter_tts: ^4.2.3
  audioplayers: ^6.5.1