Migrate loudness analyzer to audio analysis provider

[marcelveldt]

Problem

Volume-loudness detection was the last analyzer still hand-rolled in the streams and music controllers, storing measurements in its own loudness_measurements table. Every other analyzer already uses the AudioAnalysisProvider pattern. Keeping loudness separate meant duplicated plumbing, a bespoke DB table, and no reuse of the PCM chunk fan-out the new controller already does.

Changes

• New builtin loudness_analysis audio analysis provider (always enabled) - ebur128 on live PCM via the shared audio-analysis fan-out
• Removes attach_loudness_analyzer from the streams controller and set_loudness / get_loudness from the music controller; filesystem_local and opensubsonic now use mass.streams.audio_analysis.set_track_loudness(...)
• Adds loudness_album field to AudioAnalysisData so album-level loudness from tags survives the unified format
• DB migration 38→39: copies valid rows (> -50 LUFS) from loudness_measurements into audio_analysis under aa_provider_domain='loudness_analysis'; drops the old table
• Loudness hydration on streamdetails moved to just-in-time in get_queue_item_stream, so a measurement completed during a previous play is picked up without restart
• Provider skips analysis when streamdetails.volume_normalization_mode == DISABLED (respects per-player opt-out); the base-class version gating already skips tracks that already have a measurement
• Optional nightly background task (default on): batches 250 local-filesystem tracks without a loudness row, runs ebur128 directly on the file path, 2s spacing, aborts if the storage provider goes offline mid-run
• Opt-in write_replaygain_tags config (default off): writes REPLAYGAIN_TRACK_GAIN back to the file (ID3/MP4/Vorbis/APEv2) via a new write_replaygain_track_gain mutagen helper

[github-actions]

🔒 Dependency Security Report

✅ No dependency changes detected in this PR.

[Copilot]

Pull request overview

This PR migrates loudness measurement into the existing AudioAnalysisProvider framework, replacing bespoke loudness plumbing with a unified audio-analysis storage and runtime pipeline.

Changes:

• Introduces a new builtin loudness_analysis audio analysis provider (FFmpeg ebur128) and routes tag/ReplayGain-derived loudness into streams.audio_analysis.
• Consolidates loudness persistence into the audio_analysis table (schema 38→39 migration + removal of the legacy loudness_measurements table).
• Adds loudness_album to AudioAnalysisData and adds optional ReplayGain tag writing support via helpers.tags.write_replaygain_track_gain.

Reviewed changes

Copilot reviewed 12 out of 12 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
music_assistant/providers/opensubsonic/sonic_provider.py	Stores tag/ReplayGain loudness via streams.audio_analysis.set_track_loudness.
music_assistant/providers/filesystem_local/init.py	Stores parsed loudness tags via streams.audio_analysis.set_track_loudness for track/audiobook/podcast.
music_assistant/providers/loudness_analysis/provider.py	New provider implementing live PCM loudness analysis + nightly background file analysis.
music_assistant/providers/loudness_analysis/init.py	Provider setup + config entries for background analysis and ReplayGain tag writing.
music_assistant/providers/loudness_analysis/manifest.json	Declares the new builtin audio analysis provider.
music_assistant/models/audio_analysis.py	Adds loudness_album to the shared analysis data model.
music_assistant/helpers/tags.py	Adds mutagen-based helper to write REPLAYGAIN_TRACK_GAIN back to files.
music_assistant/controllers/streams/audio_buffer.py	Switches from the legacy loudness attachment to unified audio analysis startup.
music_assistant/controllers/streams/audio_analysis.py	Adds set_track_loudness helper to persist loudness into audio_analysis.
music_assistant/controllers/streams/audio.py	Moves loudness hydration to just-in-time lookup from audio_analysis in get_queue_item_stream.
music_assistant/controllers/music.py	Bumps DB schema to 39 and migrates loudness rows into audio_analysis, then drops the legacy table.
music_assistant/controllers/media/base.py	Updates library removal cleanup to delete audio analysis rows (instead of legacy loudness rows).

[Copilot]

Pull request overview

Copilot reviewed 15 out of 15 changed files in this pull request and generated no new comments.

[Copilot]

Pull request overview

Copilot reviewed 15 out of 15 changed files in this pull request and generated 1 comment.

[Copilot]

[PROBLEM] The nightly background scan iterates every available AudioAnalysisProvider, but most providers don’t override analyze_file (default returns None), so this will still fetch streamdetails for up to 250 tracks per provider and do no work, causing unnecessary DB/IO load each night.
Suggested fix: skip providers that don’t support file-based analysis (e.g., only include providers where provider.__class__.analyze_file is not AudioAnalysisProvider.analyze_file, or add an explicit capability flag/property and filter on it) before calling _find_tracks_missing_analysis/get_stream_details.