Machine Learning integration for Flutter using TensorFlow Lite and Firebase ML
日本語の概要は準備中です。原文の説明を表示しています。
Video and audio processing capabilities for Flutter applications
インストールする前に、エージェントに与えられる指示の中身を確認できます。
This skill provides comprehensive video and audio processing capabilities for Flutter applications. Includes video playback, recording, audio capture, processing, and streaming.
dependencies:
video_player: ^2.8.1
chewie: ^1.7.1 # Better video player UI
camera: ^0.10.5+5
audio_session: ^0.1.18
just_audio: ^0.9.36
record: ^5.0.4
ffmpeg_kit_flutter: ^6.0.3
image_picker: ^1.0.7
file_picker: ^6.1.1
path_provider: ^2.1.1
import 'package:video_player/video_player.dart';
import 'package:flutter/material.dart';
class VideoPlayerScreen extends StatefulWidget {
final String videoUrl;
const VideoPlayerScreen({super.key, required this.videoUrl});
@override
State<VideoPlayerScreen> createState() => _VideoPlayerScreenState();
}
class _VideoPlayerScreenState extends State<VideoPlayerScreen> {
late VideoPlayerController _controller;
@override
void initState() {
super.initState();
_controller = VideoPlayerController.network(widget.videoUrl)
..initialize().then((_) {
setState(() {});
_controller.play();
});
}
@override
Widget build(BuildContext context) {
return Scaffold(
body: Center(
child: _controller.value.isInitialized
? AspectRatio(
aspectRatio: _controller.value.aspectRatio,
child: VideoPlayer(_controller),
)
: const CircularProgressIndicator(),
),
floatingActionButton: FloatingActionButton(
onPressed: () {
setState(() {
_controller.value.isPlaying
? _controller.pause()
: _controller.play();
});
},
child: Icon(
_controller.value.isPlaying ? Icons.pause : Icons.play_arrow,
),
),
);
}
@override
void dispose() {
_controller.dispose();
super.dispose();
}
}
import 'package:chewie/chewie.dart';
class EnhancedVideoPlayer extends StatefulWidget {
final String videoUrl;
const EnhancedVideoPlayer({super.key, required this.videoUrl});
@override
State<EnhancedVideoPlayer> createState() => _EnhancedVideoPlayerState();
}
class _EnhancedVideoPlayerState extends State<EnhancedVideoPlayer> {
VideoPlayerController? _videoController;
ChewieController? _chewieController;
@override
void initState() {
super.initState();
_initializePlayer();
}
Future<void> _initializePlayer() async {
_videoController = VideoPlayerController.network(widget.videoUrl);
await _videoController!.initialize();
_chewieController = ChewieController(
videoPlayerController: _videoController!,
autoPlay: true,
looping: false,
aspectRatio: _videoController!.value.aspectRatio,
allowFullScreen: true,
allowMuting: true,
showControls: true,
placeholder: Container(
color: Colors.black,
child: const Center(
child: CircularProgressIndicator(),
),
),
);
setState(() {});
}
@override
Widget build(BuildContext context) {
return _chewieController != null
? Chewie(controller: _chewieController!)
: const Center(child: CircularProgressIndicator());
}
@override
void dispose() {
_videoController?.dispose();
_chewieController?.dispose();
super.dispose();
}
}
import 'package:camera/camera.dart';
class VideoRecorderScreen extends StatefulWidget {
const VideoRecorderScreen({super.key});
@override
State<VideoRecorderScreen> createState() => _VideoRecorderScreenState();
}
class _VideoRecorderScreenState extends State<VideoRecorderScreen> {
CameraController? _controller;
bool _isRecording = false;
String? _videoPath;
@override
void initState() {
super.initState();
_initializeCamera();
}
Future<void> _initializeCamera() async {
final cameras = await availableCameras();
if (cameras.isEmpty) return;
_controller = CameraController(
cameras.first,
ResolutionPreset.high,
enableAudio: true,
);
await _controller!.initialize();
setState(() {});
}
Future<void> _startRecording() async {
if (_controller == null || !_controller!.value.isInitialized) return;
try {
await _controller!.startVideoRecording();
setState(() => _isRecording = true);
} catch (e) {
print('Error starting recording: $e');
}
}
Future<void> _stopRecording() async {
if (!_isRecording) return;
try {
final file = await _controller!.stopVideoRecording();
setState(() {
_isRecording = false;
_videoPath = file.path;
});
} catch (e) {
print('Error stopping recording: $e');
}
}
@override
Widget build(BuildContext context) {
if (_controller == null || !_controller!.value.isInitialized) {
return const Center(child: CircularProgressIndicator());
}
return Scaffold(
body: CameraPreview(_controller!),
floatingActionButton: FloatingActionButton(
onPressed: _isRecording ? _stopRecording : _startRecording,
backgroundColor: _isRecording ? Colors.red : Colors.blue,
child: Icon(_isRecording ? Icons.stop : Icons.videocam),
),
);
}
@override
void dispose() {
_controller?.dispose();
super.dispose();
}
}
import 'package:record/record.dart';
import 'package:just_audio/just_audio.dart';
import 'package:path_provider/path_provider.dart';
import 'dart:io';
class AudioRecorderService {
final AudioRecorder _audioRecorder = AudioRecorder();
final AudioPlayer _audioPlayer = AudioPlayer();
String? _currentRecordingPath;
Future<void> startRecording() async {
final hasPermission = await _audioRecorder.hasPermission();
if (!hasPermission) {
throw Exception('Microphone permission not granted');
}
final directory = await getApplicationDocumentsDirectory();
_currentRecordingPath = '${directory.path}/recording_${DateTime.now().millisecondsSinceEpoch}.m4a';
await _audioRecorder.start(
const RecordConfig(),
path: _currentRecordingPath!,
);
}
Future<String?> stopRecording() async {
await _audioRecorder.stop();
return _currentRecordingPath;
}
Future<void> playRecording(String path) async {
await _audioPlayer.setFilePath(path);
await _audioPlayer.play();
}
Future<void> stopPlayback() async {
await _audioPlayer.stop();
}
Stream<Amplitude> get amplitudeStream => _audioRecorder.onAmplitude;
}
class AudioRecorderWidget extends ConsumerStatefulWidget {
const AudioRecorderWidget({super.key});
@override
ConsumerState<AudioRecorderWidget> createState() => _AudioRecorderWidgetState();
}
class _AudioRecorderWidgetState extends ConsumerState<AudioRecorderWidget> {
final AudioRecorderService _service = AudioRecorderService();
bool _isRecording = false;
bool _isPlaying = false;
String? _recordingPath;
double _amplitude = 0.0;
@override
Widget build(BuildContext context) {
return Column(
children: [
// Amplitude visualization
Container(
height: 100,
width: double.infinity,
color: Colors.grey[200],
child: CustomPaint(
painter: AmplitudePainter(amplitude: _amplitude),
),
),
const SizedBox(height: 20),
Row(
mainAxisAlignment: MainAxisAlignment.center,
children: [
// Record button
FloatingActionButton(
onPressed: _toggleRecording,
backgroundColor: _isRecording ? Colors.red : Colors.blue,
child: Icon(_isRecording ? Icons.stop : Icons.mic),
),
const SizedBox(width: 20),
// Play button
if (_recordingPath != null)
FloatingActionButton(
onPressed: _togglePlayback,
backgroundColor: _isPlaying ? Colors.orange : Colors.green,
child: Icon(_isPlaying ? Icons.stop : Icons.play_arrow),
),
],
),
],
);
}
Future<void> _toggleRecording() async {
if (_isRecording) {
final path = await _service.stopRecording();
setState(() {
_isRecording = false;
_recordingPath = path;
});
} else {
await _service.startRecording();
setState(() => _isRecording = true);
// Listen to amplitude
_service.amplitudeStream.listen((amp) {
setState(() => _amplitude = amp.current);
});
}
}
Future<void> _togglePlayback() async {
if (_isPlaying) {
await _service.stopPlayback();
setState(() => _isPlaying = false);
} else if (_recordingPath != null) {
await _service.playRecording(_recordingPath!);
setState(() => _isPlaying = true);
}
}
}
// Custom painter for amplitude visualization
class AmplitudePainter extends CustomPainter {
final double amplitude;
AmplitudePainter({required this.amplitude});
@override
void paint(Canvas canvas, Size size) {
final paint = Paint()
..color = Colors.blue
..strokeWidth = 2;
final height = size.height;
final centerY = height / 2;
final barWidth = size.width / 50;
for (int i = 0; i < 50; i++) {
final barHeight = (amplitude / 100) * height * (0.5 + (i % 5) * 0.1);
final x = i * barWidth;
canvas.drawLine(
Offset(x, centerY - barHeight / 2),
Offset(x, centerY + barHeight / 2),
paint,
);
}
}
@override
bool shouldRepaint(covariant CustomPainter oldDelegate) => true;
}
import 'package:ffmpeg_kit_flutter/ffmpeg_kit.dart';
import 'package:ffmpeg_kit_flutter/return_code.dart';
class VideoProcessingService {
/// Trim video to specific duration
static Future<String?> trimVideo(
String inputPath,
String outputPath,
Duration start,
Duration duration,
) async {
final command =
'-i "$inputPath" -ss ${start.inSeconds} -t ${duration.inSeconds} '
'-c copy "$outputPath"';
final session = await FFmpegKit.execute(command);
final returnCode = await session.getReturnCode();
if (ReturnCode.isSuccess(returnCode)) {
return outputPath;
}
return null;
}
/// Compress video
static Future<String?> compressVideo(
String inputPath,
String outputPath, {
String quality = 'medium', // low, medium, high
}) async {
String crf = '23'; // Default medium
if (quality == 'low') crf = '28';
if (quality == 'high') crf = '18';
final command =
'-i "$inputPath" -vcodec h264 -acodec aac -crf $crf "$outputPath"';
final session = await FFmpegKit.execute(command);
final returnCode = await session.getReturnCode();
return ReturnCode.isSuccess(returnCode) ? outputPath : null;
}
/// Extract thumbnail from video
static Future<String?> extractThumbnail(
String videoPath,
String outputPath, {
Duration position = const Duration(seconds: 1),
}) async {
final command =
'-i "$videoPath" -ss ${position.inSeconds} -vframes 1 "$outputPath"';
final session = await FFmpegKit.execute(command);
final returnCode = await session.getReturnCode();
return ReturnCode.isSuccess(returnCode) ? outputPath : null;
}
/// Get video information
static Future<Map<String, dynamic>> getVideoInfo(String videoPath) async {
final session = await FFmpegKit.execute('-i "$videoPath"');
final output = await session.getOutput();
final logs = await session.getLogs();
// Parse duration, resolution, etc. from output
final info = <String, dynamic>{};
for (final log in logs) {
final message = log.getMessage();
if (message?.contains('Duration') == true) {
final match = RegExp(r'Duration: (\d{2}):(\d{2}):(\d{2}\.\d{2})')
.firstMatch(message!);
if (match != null) {
final hours = int.parse(match.group(1)!);
final minutes = int.parse(match.group(2)!);
final seconds = double.parse(match.group(3)!);
info['duration'] = Duration(
hours: hours,
minutes: minutes,
seconds: seconds.toInt(),
);
}
}
}
return info;
}
}
まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Machine Learning integration for Flutter using TensorFlow Lite and Firebase ML
日本語の概要は準備中です。原文の説明を表示しています。
Google Maps and Mapbox integration for Flutter applications
日本語の概要は準備中です。原文の説明を表示しています。
Implement tasks from an OpenSpec change. Use when the user wants to start implementing, continue implementation, or work through tasks.
日本語の概要は準備中です。原文の説明を表示しています。
Archive a completed change in the experimental workflow. Use when the user wants to finalize and archive a change after implementation is complete.
日本語の概要は準備中です。原文の説明を表示しています。
Enter explore mode - a thinking partner for exploring ideas, investigating problems, and clarifying requirements. Use when the user wants to think through something before or during a change.
日本語の概要は準備中です。原文の説明を表示しています。
Propose a new change with all artifacts generated in one step. Use when the user wants to quickly describe what they want to build and get a complete proposal with design, specs, and tasks ready for implementation.
日本語の概要は準備中です。原文の説明を表示しています。