AudioDB

Overview & Philosophy

AudioDB is designed for mixed collections (audio, MIDI, tracker, video clips), inconsistent tags and need for fast operative over large dataset. The primary goal is to have one stable desktop tool that manages ingest, analysis, metadata validation, export and practical "copy workflow" for curatorial use.

Use- case orientation

• Archiving: cataloguing large collections with strong quality variability tags.
• Curitarianism: fast filtering according to BPM, key, year, genre and custom fields.
• Integration: export to XLSX / CSV / HTML and follow-up work in other tools.
• Workflow "I Want This": staging + transfer of selected subset to the target medium.

Architectural principles

• core: minimum runtime, fast start, high compatibility.
• TkinterDnD2: drag & drop flow for files and folders without hand browse step.
• Subprocess analysis: Heavy calculations run outside the graphical interface process.
• Export-first model: XLSX / CSV / HTML are considered as the primary integration interface.
• Session playback: the playback is separated from DB writing to avoid the risk of blocking the UI.

Supported formats

The default set of connections is designed as a wide "ingest minimum" for real collections, where modern audio, historical formats and video clips are mixed as musical references.

• Audio:
• MIDI:
• SID:
• Tracker:
• Video:

Practical compatibility

• Tracker / SID files are fully registered in database and export.
• Playback and detailed analysis of the tracker / SID may require an external backend tool.
• Video formats are used primarily for recording, preview and metadata workflow.
• The list extension is possible via filter application settings (scan include / extensions).

In-service format strategy

1. First, run a QUICK scan through the entire rota for inventory.
2. DEEP / FULL is available except for selected subsets.
3. Save the unknown or problematic extension in a custom flag for later audit.

I want a workflow

Workflow is designed for logistics of selection of tracks to the target folder (USB, staging, release subset). In practice, two schemes are available.

09.1 Direct copying

• Action: action I want that.
• Application: quick one-off action without intermediate action.
• Advantage: minimum number of clicks.
• Risk: weaker control of target structure before launching.

09.2 Staging pipeline

1. addition to continuous selection - continuous compilation of selection across filters.
2. preview and validation of sources and targets - validation of sources / targets before copying.
3. final creation of transmission task - final job and transmission.

09.3 Recommendations for large copies

• On Windows use robocopy mode if available in configuration.
• Always keep the log summary: number of copied files, errors, missing files.
• Work with the manifesto and timestamp export on repeated packaging.

Exports + streaming server + subtitles

10.1 Exports

• XLSX: ideal for manual editing and rundtrip import.
• CSV: suitable for automation, diffy and BI pipeline.
• HTML: reports and playlists; for robust viewing is available .

10.2 Streaming server (Range Requests)

implements HTTP Range Requests for instant session / playback of large files without full download. It is designed for localhost / LAN scenarios and is especially suitable for HTML exports with audio links.

• support RFC 7233 range
• single-third model
• Graceful shutdown
• emphasis on local traffic, not public hosting

10.3 Subtitles and text accompanying files

AudioDB does not solve the full subtitle pipeline as the primary function of the player in this version; it works mainly with metadata and exports. If there are external text escorts (lyrics / notes) in the library, the recommended procedure is to register them via custom field or subsequent export workflow.

• Embedded subtitle stream is not the main target scenario of AudioDB monolith
• external text files can be mapped over filepath / custom columns
• for advanced playback scenarios it is advisable to connect to a specialized player

Troubleshooting + diagnostics

14.1 Quick problem matrix

• Deep / Full scan fails: Authentication /, run a worker test, fall back on Quick.
• Lookup bugs: check tokens, internet, rate-limit and retry interval.
• FFmpeg unavailable: Authentication and path in configuration.
• Slow AI: paging, less columns, scan dosed.
• Playback doesn't work: fallback to System player provider, check backend options.
• HTML playback not working: use or local streaming server.

14.2 Diagnostic flow

1. Repeat the problem for a little dataset test.
2. Save log export and mark active config keys.
3. Check if the problem arises after the configuration reset.
4. Only then do the reindex all or the migration DB.

14.3 Known limits (runtime)

• Mastering-grade loutness and true-peak metrics are not the target of the application.
• The visualizer is not intended as a DAW realtime DSP tool.
• Multimachine DB sync is dealt with export / import workflow, not distributed DB.

14.4 FAQ (operational issues)

Why does a portion of the results break after the library is moved?

The primary identity of the line is . If the root path changes, the old records need to be reindexed or migrate by mapping the paths. Hash helps with the deduce, but does not replace the change of absolute path.

Is Deep scan better at all?

No. For stability, dosing after ingredients or profiles is better. Large disposable runs increase the risk of a timeout worker and complicate recovery.

When to use Save DB As instead of Save DB?

In larger mass operations (mass import, global editing, road map) always use action Save Database as... And work on your snapshot.

How to solve slow browsing table?

Limit visible column, hold paging, do not use show all lines and check whether there is a challenging scan or a look at the same time.

Can I share HTML export publicly?

Yes, only after anonymization. Export may contain local routes or internal identifiers. Before publishing, remove sensitive data.