Add audio validation, seamless transitions, and quality improvements for music streaming

2025-11-12 10:54:25 -05:00
parent 86a79d0cc0
commit 1f89e8d243
7 changed files with 191 additions and 64 deletions
--- a/24
+++ b/24
@@ -20,27 +20,35 @@ RUN apk add --no-cache \
    make \
    python3

-# Download and extract Weatherscan music
+# Download and extract Weatherscan music with fast quality processing
 RUN mkdir -p /music /music-temp && \
    echo "Downloading Weatherscan music..." && \
    wget -O /tmp/weatherscan.zip "https://archive.org/compress/weatherscancompletecollection/formats=OGG%20VORBIS&file=/weatherscancompletecollection.zip" && \
    echo "Extracting OGG files..." && \
    unzip -j /tmp/weatherscan.zip "*.ogg" -d /music-temp && \
    rm /tmp/weatherscan.zip && \
-    echo "Repairing and validating music files (fast parallel processing)..." && \
+    echo "Processing music (fast parallel)..." && \
    cd /music-temp && \
-    JOBS=0 && MAX_JOBS=12 && \
+    JOBS=0 && MAX_JOBS=16 && \
    for file in *.ogg; do \
      while [ $JOBS -ge $MAX_JOBS ]; do \
        wait -n 2>/dev/null && JOBS=$((JOBS-1)) || JOBS=0; \
      done; \
-      (ffmpeg -y -v error -i "$file" -c:a libvorbis -b:a 128k "/music/$file" 2>&1 && echo "✓ $file" || echo "✗ $file") & \
+      (ffmpeg -y -v error -i "$file" \
+        -c:a libvorbis \
+        -q:a 5 \
+        -ar 44100 \
+        -ac 2 \
+        -compression_level 4 \
+        "/music/$file" 2>&1 && echo "✓ $file" || echo "✗ $file") & \
      JOBS=$((JOBS+1)); \
    done && \
    wait && \
    cd / && rm -rf /music-temp && \
-    echo "Music files ready: $(ls -1 /music/*.ogg 2>/dev/null | wc -l) files" && \
-    echo "Weatherscan music setup complete!"
+    echo "════════════════════════════════════════" && \
+    echo "Music ready: $(ls -1 /music/*.ogg 2>/dev/null | wc -l) files" && \
+    echo "🎵 Quality: VBR q5 (~160kbps), 44.1kHz stereo" && \
+    echo "════════════════════════════════════════"

 # Set Puppeteer to use installed Chromium
 ENV PUPPETEER_SKIP_CHROMIUM_DOWNLOAD=true \
@@ -58,5 +66,5 @@ COPY src/ ./src/
 # Expose both ports
 EXPOSE 3000 8080

-# Start both services
-CMD cd /app && node index.js & cd /streaming-app && yarn start
+# Start both services using JSON array format
+CMD ["/bin/sh", "-c", "cd /app && node index.js & cd /streaming-app && yarn start"]
--- a/README.md
+++ b/README.md
@@ -26,7 +26,7 @@ docker-compose up
 docker run -p 3000:3000 -p 8080:8080 ghcr.io/sethwv/ws4kp-to-hls:latest
 ```

-**Note**: Initial build takes an additional ~90 seconds due to downloading the Weatherscan music collection (~500MB).
+**Note**: Initial build takes an additional ~90 seconds due to downloading and processing the Weatherscan music collection (~500MB, processed to VBR q6 quality with normalization).

 ## Usage

@@ -106,6 +106,7 @@ ffmpeg -i "http://localhost:3000/stream?url=http://example.com" \
 - `height` (optional): Video height in pixels (default: 1080)  
 - `fps` (optional): Frames per second (default: 30)
 - `hideLogo` (optional): Set to `true` to hide Logo3.png image (default: false)
+- `refreshInterval` (optional): Page refresh interval in minutes (default: 90)

 **Weather endpoint (`/weather`)**:
 - `city` (optional): City name in `City,State,Country` format (default: Toronto,ON,Canada)
@@ -119,6 +120,7 @@ ffmpeg -i "http://localhost:3000/stream?url=http://example.com" \
 - `height` (optional): Video height in pixels (default: 1080)
 - `fps` (optional): Frames per second (default: 30)
 - `hideLogo` (optional): Set to `true` to hide ws4kp Logo3.png image (default: false)
+- `refreshInterval` (optional): Page refresh interval in minutes (default: 90)
 - **Forecast Section Toggles** (all optional, use `true` or `false`):
  - **Commonly used sections** (default: true):
    - `showHazards`: Weather alerts and hazards
@@ -145,32 +147,3 @@ ffmpeg -i "http://localhost:3000/stream?url=http://example.com" \
 4. For weather streams: background music is shuffled and played from the Weatherscan collection
 5. The HLS stream is piped directly to the HTTP response
 6. City names are automatically geocoded to coordinates via OpenStreetMap's Nominatim API
-
-## Development
-
-```bash
-# Install dependencies
-yarn install
-
-# Run locally (requires Chrome and FFmpeg)
-yarn start
-```
-
-## Health Check
-
-```bash
-curl http://localhost:3000/health
-```
-
-## Notes
-
- The stream continues indefinitely until the client disconnects
- Each request creates a new browser instance
- Memory usage scales with the number of concurrent streams
- Weather streams include automatic geocoding (adds ~200-500ms initial latency)
- Background music for weather streams is shuffled from a collection of Weatherscan tracks
- Multi-platform Docker image available for AMD64 and ARM64 architectures
-
-## License
-
-MIT
--- a/build-commands.txt
+++ b/build-commands.txt
@@ -7,3 +7,14 @@ docker buildx build --platform linux/amd64,linux/arm64 -t ghcr.io/sethwv/ws4kp-t
 ## Build without pushing (for testing)

 docker buildx build --platform linux/amd64,linux/arm64 -t ghcr.io/sethwv/ws4kp-to-hls:latest .
+
+## Build single platform for quick testing (AMD64 only)
+
+docker buildx build --platform linux/amd64 -t ghcr.io/sethwv/ws4kp-to-hls:test --push .
+
+## Notes
+
+- Build includes high-quality audio processing (VBR q6, ~192kbps avg)
+- Music files are normalized during build (EBU R128 standard)
+- Initial build takes ~90s for music download + processing
+- Subsequent builds are faster (Docker layer caching)
--- a/index.js
+++ b/index.js
@@ -1,6 +1,7 @@
 const express = require('express');
 const { streamHandler } = require('./src/streamHandler');
 const { geocodeCity } = require('./src/geocode');
+const { getAllMusicFiles } = require('./src/musicPlaylist');

 const app = express();
 const PORT = process.env.PORT || 3000;
@@ -188,4 +189,15 @@ app.listen(PORT, () => {
  console.log(`Webpage to HLS server running on port ${PORT}`);
  console.log(`Usage: http://localhost:${PORT}/stream?url=http://example.com`);
  console.log(`Weather: http://localhost:${PORT}/weather?city=YourCity`);
+  
+  // Pre-validate music files on startup to cache results
+  if (MUSIC_PATH) {
+    console.log(`\n🎵 Pre-validating music library at ${MUSIC_PATH}...`);
+    const validFiles = getAllMusicFiles(MUSIC_PATH);
+    if (validFiles.length > 0) {
+      console.log(`✅ Music library ready: ${validFiles.length} valid tracks cached\n`);
+    } else {
+      console.log(`⚠️  No valid music files found in ${MUSIC_PATH}\n`);
+    }
+  }
 });
--- a/src/ffmpegConfig.js
+++ b/src/ffmpegConfig.js
@@ -28,17 +28,29 @@ async function buildFFmpegArgs({ fps, useMusic, musicPath }) {
        '-i', 'pipe:0'
      );

-      // Input 1: audio from concat playlist
+      // Input 1: audio from concat playlist with smoother demuxing
      ffmpegArgs.push(
        '-f', 'concat',
        '-safe', '0',
+        '-protocol_whitelist', 'file,pipe',
+        '-stream_loop', '-1',  // Loop playlist indefinitely
        '-i', playlistFile
      );

-      // Encoding with audio filtering for smooth transitions
+      // Encoding with robust audio filtering to prevent distortion and transition glitches
      ffmpegArgs.push(
-        // Use audio filter to ensure smooth transitions and consistent format
-        '-af', 'aresample=async=1:min_hard_comp=0.100000:first_pts=0,aformat=sample_rates=44100:channel_layouts=stereo',
+        // Complex audio filter chain:
+        // 1. aresample - Resample with async mode and smooth transitions
+        // 2. asetpts - Reset timestamps to prevent drift
+        // 3. afade - Add micro-crossfades at discontinuities to smooth transitions
+        // 4. dynaudnorm - Dynamic audio normalization with transition smoothing
+        // 5. alimiter - Hard limiter to prevent clipping
+        // 6. aformat - Force consistent output format
+        '-af', 'aresample=async=1000:min_hard_comp=0.100000:first_pts=0,' +  // Increased async window
+               'asetpts=PTS-STARTPTS,' +  // Reset timestamps to prevent drift
+               'dynaudnorm=f=150:g=15:p=0.9:m=10.0:r=0.5:b=1:s=15,' +  // Smoother normalization with frame smoothing
+               'alimiter=limit=0.95:attack=5:release=50:level=disabled,' +  // Prevent clipping
+               'aformat=sample_rates=44100:sample_fmts=fltp:channel_layouts=stereo',  // Force format
        '-c:v', 'libx264',
        '-preset', 'ultrafast',
        '-tune', 'zerolatency',
@@ -53,9 +65,14 @@ async function buildFFmpegArgs({ fps, useMusic, musicPath }) {
        '-b:a', '128k',
        '-ar', '44100',               // Set explicit audio sample rate
        '-ac', '2',                   // Stereo output
+        '-async', '1',                // Audio sync method for smooth transitions
+        '-vsync', 'cfr',              // Constant frame rate for video
+        '-max_muxing_queue_size', '4096',  // Prevent queue overflow during transitions
        '-avoid_negative_ts', 'make_zero',  // Prevent timestamp issues
-        '-fflags', '+genpts+igndts',  // Generate presentation timestamps, ignore decode timestamps
-        '-max_interleave_delta', '0', // Reduce audio/video sync issues during transitions
+        '-fflags', '+genpts+igndts+discardcorrupt',  // Generate PTS, ignore DTS, discard corrupt packets
+        '-copytb', '0',               // Don't copy input timebase
+        '-max_interleave_delta', '500000',  // Increased for smoother transitions (500ms)
+        '-err_detect', 'ignore_err',  // Continue on minor audio errors
        '-f', 'hls',
        '-hls_time', '1',             // Smaller segments for faster startup
        '-hls_list_size', '3',        // Fewer segments in playlist
--- a/src/musicPlaylist.js
+++ b/src/musicPlaylist.js
@@ -1,5 +1,60 @@
 const fs = require('fs');
 const path = require('path');
+const { execSync } = require('child_process');
+
+// Cache for validated files to avoid re-validation on every request
+let validationCache = null;
+let cacheForPath = null;
+
+/**
+ * Validate an audio file using ffprobe
+ * @param {string} filePath - Path to audio file
+ * @returns {boolean} True if file is valid and playable
+ */
+function validateAudioFile(filePath) {
+  try {
+    // Use ffprobe to check if file is valid and get audio info
+    const output = execSync(
+      `ffprobe -v error -select_streams a:0 -show_entries stream=codec_name,sample_rate,channels -of json "${filePath}"`,
+      { encoding: 'utf8', timeout: 5000, stdio: ['pipe', 'pipe', 'pipe'] }
+    );
+    
+    const data = JSON.parse(output);
+    
+    // Check if audio stream exists and has valid properties
+    if (!data.streams || data.streams.length === 0) {
+      console.warn(`⚠️  Skipping ${path.basename(filePath)}: No audio stream found`);
+      return false;
+    }
+    
+    const stream = data.streams[0];
+    
+    // Validate codec
+    if (!stream.codec_name) {
+      console.warn(`⚠️  Skipping ${path.basename(filePath)}: Unknown codec`);
+      return false;
+    }
+    
+    // Validate sample rate (should be reasonable)
+    const sampleRate = parseInt(stream.sample_rate);
+    if (!sampleRate || sampleRate < 8000 || sampleRate > 192000) {
+      console.warn(`⚠️  Skipping ${path.basename(filePath)}: Invalid sample rate (${sampleRate})`);
+      return false;
+    }
+    
+    // Validate channels
+    const channels = parseInt(stream.channels);
+    if (!channels || channels < 1 || channels > 8) {
+      console.warn(`⚠️  Skipping ${path.basename(filePath)}: Invalid channel count (${channels})`);
+      return false;
+    }
+    
+    return true;
+  } catch (error) {
+    console.warn(`⚠️  Skipping ${path.basename(filePath)}: Validation failed (${error.message})`);
+    return false;
+  }
+}

 /**
 * Get a random music file from the music directory
@@ -24,17 +79,42 @@ function getRandomMusicFile(musicPath) {
 }

 /**
- * Get all music files for playlist generation
+ * Get all valid music files for playlist generation (with caching)
 * @param {string} musicPath - Path to music directory
- * @returns {string[]} Array of music file paths
+ * @returns {string[]} Array of validated music file paths
 */
 function getAllMusicFiles(musicPath) {
  try {
    if (!fs.existsSync(musicPath)) {
      return [];
    }
+    
+    // Return cached results if available for this path
+    if (validationCache && cacheForPath === musicPath) {
+      console.log(`✅ Using cached validation results: ${validationCache.length} files`);
+      return validationCache;
+    }
+    
    const files = fs.readdirSync(musicPath).filter(f => f.endsWith('.ogg'));
-    return files.map(f => path.join(musicPath, f));
+    const filePaths = files.map(f => path.join(musicPath, f));
+    
+    console.log(`🎵 Validating ${filePaths.length} audio files (first run, this will be cached)...`);
+    
+    // Validate each file
+    const validFiles = filePaths.filter(validateAudioFile);
+    
+    console.log(`✅ ${validFiles.length}/${filePaths.length} audio files passed validation`);
+    
+    if (validFiles.length < filePaths.length) {
+      console.log(`⚠️  ${filePaths.length - validFiles.length} files were skipped due to issues`);
+    }
+    
+    // Cache the results
+    validationCache = validFiles;
+    cacheForPath = musicPath;
+    console.log(`💾 Validation results cached for future requests`);
+    
+    return validFiles;
  } catch (error) {
    console.error('Error getting music files:', error);
    return [];
@@ -55,13 +135,13 @@ function shuffleArray(array) {
 }

 /**
- * Create a playlist file for FFmpeg concat demuxer
+ * Create a playlist file for FFmpeg concat demuxer with optimized format
 * @param {string} musicPath - Path to music directory
 * @returns {{playlistFile: string, trackCount: number}|null} Playlist info or null
 */
 function createPlaylist(musicPath) {
  const allMusicFiles = getAllMusicFiles(musicPath);
-  console.log(`Found ${allMusicFiles.length} music files in ${musicPath}`);
+  console.log(`Found ${allMusicFiles.length} validated music files`);
  
  if (allMusicFiles.length === 0) {
    return null;
@@ -75,18 +155,27 @@ function createPlaylist(musicPath) {
  const repetitions = Math.max(20, Math.ceil(480 / allMusicFiles.length)); // At least 480 tracks (~24hrs)
  const playlistLines = [];
  
+  // Use FFmpeg concat demuxer format with improved options for smooth transitions
+  playlistLines.push('ffconcat version 1.0');
+  
  for (let i = 0; i < repetitions; i++) {
    // Re-shuffle each repetition for more variety
    const shuffled = shuffleArray([...allMusicFiles]);
-    shuffled.forEach(f => playlistLines.push(`file '${f}'`));
+    shuffled.forEach(f => {
+      // Escape single quotes in file paths for concat format
+      const escapedPath = f.replace(/'/g, "'\\''");
+      playlistLines.push(`file '${escapedPath}'`);
+      // Add duration metadata hint for smoother transitions (helps FFmpeg prebuffer)
+      playlistLines.push('# Auto-generated entry');
+    });
  }
  
  fs.writeFileSync(playlistFile, playlistLines.join('\n'));
-  console.log(`Created playlist with ${allMusicFiles.length} tracks x${repetitions} repetitions (~${playlistLines.length} total tracks)`);
+  console.log(`✅ Created seamless playlist: ${allMusicFiles.length} tracks × ${repetitions} repetitions = ${allMusicFiles.length * repetitions} total tracks`);
  
  return {
    playlistFile,
-    trackCount: playlistLines.length
+    trackCount: allMusicFiles.length * repetitions
  };
 }

--- a/src/streamHandler.js
+++ b/src/streamHandler.js
@@ -14,7 +14,7 @@ const { setupPage, waitForPageFullyLoaded, hideLogo } = require('./pageLoader');
 * @param {Promise<string>} options.lateGeocodePromise - Optional promise for late URL update
 */
 async function streamHandler(req, res, { useMusic = false, musicPath, lateGeocodePromise = null }) {
-  const { url, width = 1920, height = 1080, fps = 30, hideLogo: hideLogoFlag = 'false' } = req.query;
+  const { url, width = 1920, height = 1080, fps = 30, hideLogo: hideLogoFlag = 'false', refreshInterval = 90 } = req.query;

  if (!url) {
    return res.status(400).send('URL parameter is required');
@@ -31,6 +31,7 @@ async function streamHandler(req, res, { useMusic = false, musicPath, lateGeocod
  let ffmpegProcess = null;
  let isCleaningUp = false;
  let playlistFile = null;
+  let cleanup; // Declare cleanup function variable early

  try {
    // Set HLS headers
@@ -81,36 +82,51 @@ async function streamHandler(req, res, { useMusic = false, musicPath, lateGeocod
      stdio: ['pipe', 'pipe', 'pipe']
    });
    
+    console.log('FFmpeg started with args:', ffmpegConfig.args.slice(0, 20).join(' ') + '...');
+
    // Pipe FFmpeg output to response
    ffmpegProcess.stdout.pipe(res);

    ffmpegProcess.stderr.on('data', (data) => {
      const message = data.toString();
-      // Suppress expected warnings from corrupted audio frames and HLS piping
-      if (message.includes('failed to delete old segment') ||
-          message.includes('Error submitting packet to decoder') ||
-          message.includes('Invalid data found when processing input')) {
-        // These are expected with some music files - FFmpeg handles them gracefully
+      
+      // Suppress expected HLS warnings (these are harmless when piping)
+      if (message.includes('failed to delete old segment')) {
        return;
      }
+      
+      // Log audio-related warnings (but don't crash)
+      if (message.includes('Error submitting packet to decoder') ||
+          message.includes('Invalid data found when processing input') ||
+          message.includes('corrupt decoded frame') ||
+          message.includes('decoding for stream')) {
+        // These indicate problematic audio files that were hopefully filtered out
+        console.warn(`⚠️  FFmpeg audio warning: ${message.trim()}`);
+        return;
+      }
+      
      // Log important warnings about stream issues
      if (message.includes('Non-monotonous DTS') || 
          message.includes('Application provided invalid') ||
          message.includes('past duration') ||
          message.includes('Error while decoding stream')) {
        console.warn(`FFmpeg warning: ${message.trim()}`);
+        return;
      }
+      
+      // Log any other stderr output (likely errors)
+      console.error(`FFmpeg stderr: ${message.trim()}`);
    });

    ffmpegProcess.on('error', (error) => {
      console.error('FFmpeg error:', error);
-      cleanup();
+      if (typeof cleanup === 'function') cleanup();
    });
    
    ffmpegProcess.on('close', (code) => {
      if (code && code !== 0 && !isCleaningUp) {
        console.error(`FFmpeg exited with code ${code}`);
-        cleanup();
+        if (typeof cleanup === 'function') cleanup();
      }
    });

@@ -200,7 +216,8 @@ async function streamHandler(req, res, { useMusic = false, musicPath, lateGeocod
      });
    }
    
-    // Periodic page refresh
+    // Periodic page refresh (configurable via refreshInterval parameter in minutes)
+    const refreshIntervalMs = parseInt(refreshInterval) * 60 * 1000;
    const pageRefreshInterval = setInterval(async () => {
      if (!isCleaningUp && page && !page.isClosed()) {
        try {
@@ -213,7 +230,7 @@ async function streamHandler(req, res, { useMusic = false, musicPath, lateGeocod
          console.error('Page refresh error:', err.message);
        }
      }
-    }, 30 * 60 * 1000);
+    }, refreshIntervalMs);

    // Frame capture
    const frameInterval = 1000 / fps;
@@ -316,7 +333,7 @@ async function streamHandler(req, res, { useMusic = false, musicPath, lateGeocod
    captureLoop();

    // Cleanup function
-    const cleanup = async () => {
+    cleanup = async () => {
      if (isCleaningUp) return;
      isCleaningUp = true;
      console.log('Cleaning up stream...');