Documentation

Getting Started

SMDB Companion finds and removes duplicate audio files from your Soundminer databases — or directly from folders of audio files if you don't use Soundminer.

The short version of the workflow:

1. Open a database (or pick folders to scan)
2. Turn on the search algorithms you want
3. Search, then review the results
4. Adjust which records are checked for removal
5. Process the records

Everything below explains those steps in more depth. If you'd rather watch than read, the video overview is a good tour.

Where the settings live

Two places. The Options menu holds the switches you change between runs — safety database, what happens to files on disk, fingerprint handling, presets. Settings (Cmd/Ctrl+,) holds the five configuration tabs:

Match Criteria	What counts as a duplicate, plus Safe Folders and audio fingerprint options
Preservation Order	Which file wins when duplicates are found
Tags Manager	AudioSuite and filename tag lists
Keyboard Shortcuts	Remap any shortcut
Appearance	Colour schemes and individual colours

Installation

macOS

1. Download the .dmg from the download page
2. Open it and drag SMDB Companion to your Applications folder
3. If the app is blocked on first launch, open System Settings → Privacy & Security and allow apps from identified developers

The macOS build is a universal binary — one download runs natively on both Apple Silicon and Intel. Requires macOS 10.15 or later.

Windows

1. Download the .msi installer
2. Run it and follow the prompts
3. Launch SMDB Companion from the Start menu

Requires 64-bit Windows 10 or later.

Registration & Trial

SMDB Companion is free to try, with one limit: the trial scans but does not change anything.

Unregistered, you can open any database, run every search algorithm, and see how many records were found. The results table itself and the option to process records stay locked until you enter a license.

That's deliberate — you can measure exactly how much duplication is in your library, and how well your settings work, before deciding whether it's worth paying for.

To register, open the Registration view (5) and enter the name, email, and license key from your purchase email. The key is tied to the name and email exactly as they appear there.

Lost your key? Use license recovery — enter your purchase email and it will be sent to you.

Opening a Database

SMDB Companion works four ways, all under the File menu:

• Open Database (Cmd/Ctrl+O) — pick a Soundminer .sqlite database. The file picker opens in your Soundminer folder automatically.
• Open Recent (Cmd/Ctrl+Shift+O) — reopen a database you've used before. Entries that no longer exist are cleaned out automatically.
• Connect to Server — for Soundminer server setups. Enter username, password, address, and port, then choose a database from the list.
• File System Search (Cmd/Ctrl+F) — no Soundminer required. Add folders and SMDB Companion scans the audio files directly.

Reveal Database Folder (Shift+Cmd/Ctrl+R) opens your Soundminer database folder in Finder or Explorer. SMDB Companion locates it for whichever Soundminer version you have installed — V6, V7, or later — including the case where the V7 folder is a symlink back to V6.

The Four Views

The same session can be looked at four ways. Switch with the number keys:

• Search — choose your database and algorithms
• Results (2) — review and edit what was found
• Split (3) — search and results side by side, useful when tuning settings and re-running
• No Frills (4) — a stripped-down single-screen layout for fast repeat work

Cmd/Ctrl+, opens Settings; the , key alone toggles the toolbars if you want more room for the table.

Search View

This is where a scan is set up. It has two parts.

The database selector shows what is currently open and lets you change it.

The algorithm list is where you turn checks on and off. Each algorithm has a tooltip explaining what it does — hover before enabling something unfamiliar. Two of them, Similar Filename and Audiosuite Tags, are refinements of Duplicate Search and appear greyed out unless Duplicate Search is enabled.

Some algorithms carry their own settings inline: Minimum Duration takes a length in seconds, and Database Compare has a Select DB button for choosing what to compare against.

When you're ready, Search Database (Cmd/Ctrl+Enter) starts the scan. Progress is shown as it runs, and Esc cancels. Large libraries with audio comparison enabled can take a long time — that's expected.

Search Algorithms

Nine algorithms, each finding a different kind of problem. They can be combined freely; a record found by more than one is still only listed once, tagged with every reason it matched.

Duplicate Search

The core check. Finds records that match each other on the Match Criteria you've defined — by default Filename, Channels, and Duration. This is the one to start with.

Similar Filename

Strips trailing digits and letters that accumulate on working copies — Footsteps.1.4.21.M.wav is recognised as a variant of Footsteps.wav. Requires Duplicate Search.

Audiosuite Tags

Finds Pro Tools AudioSuite render tags in filenames and checks whether the original unprocessed file is also in your library. Requires Duplicate Search. The tag list is editable — see Tags Manager.

Audio Content Comparison

Compares the actual audio rather than the metadata, so it catches duplicates that were renamed or re-encoded. The most thorough check and by far the slowest — see Audio Fingerprints for how to make repeat runs fast.

Dual Mono Check

Finds files where every channel contains identical audio — stereo files that are really mono. These aren't duplicates of other files, so they're handled separately: in the results you can strip the redundant channels instead of removing the record.

Filename Contains Tag

Marks any file whose name contains one of your Filename Tags. Useful for sweeping out working files with a naming convention you control, like _TEMP or .bounce.

Invalid Files

Finds records pointing at files that are no longer on disk — the residue of moved or deleted media. A good first run on any database that's been in use for years.

Minimum Duration

Marks files shorter than a length you set. Catches clicks, accidental one-frame recordings, and truncated files. Default is 0.5 seconds.

Database Compare

Compares the open database against a second one and marks records that appear in both. Useful for merging libraries, or for checking a new acquisition against what you already own.

Match Criteria

Found under Settings → Match Criteria, which also holds Safe Folders and the audio fingerprint options.

Match Criteria define what "duplicate" means for Duplicate Search. Two records are duplicates only if every criterion in the list matches.

The default is Filename, Channels, and Duration — conservative, and unlikely to group files that aren't genuinely the same. You can add any database column, or remove criteria to cast a wider net.

Ignore Filetypes (extensions) makes filename comparison ignore the extension, so Thunder.wav and Thunder.aif are treated as the same name. Turn this on if your library holds the same material in several formats and you only want one.

Preservation Priority

Once a group of duplicates is found, something has to decide which one to keep. That's what Preservation Priority does, and it's the setting most worth your time — it's what makes the results match how your library is organised.

Configure it under Settings → Preservation Order.

How the rules are applied

Rules run in order, top to bottom. The first rule that distinguishes the candidates decides the winner. If they all tie on a rule, the next rule is checked. So rules at the top matter most, and you reorder them by dragging.

Each rule is a column, an operator, and (for some operators) a value:

• is / is NOT — exact match against your value
• contains / does NOT contain — substring match, the workhorse for path rules
• is empty / is NOT empty — presence of any value, good for preferring described records
• is largest / is smallest — numeric comparison for duration, sample rate, bit depth, dates

The defaults, and why

Out of the box the order is roughly: prefer records that have a description; prefer paths not in "Audio Files" (a session folder, not a library); prefer paths containing LIBRARY or LIBRARIES; then fall back to longest duration, most channels, highest sample rate, highest bit depth, and finally the oldest original and scan dates.

That encodes one idea: a catalogued library copy beats a session copy, and when both are library copies, the higher-quality file wins.

Results View

Results are grouped: each set of duplicates appears together, with the record SMDB Companion intends to keep and the ones it intends to remove colour-coded. A checked box means "process this record."

Nothing here is final. The results are a proposal. Review them and change whatever you disagree with before processing.

Filtering

The Filter by menu narrows what's on screen:

• All Records — everything in the database
• Relevant Records — only records involved in a match (the default)
• Records to Keep / Records to Remove — the two sides of the decision
• By algorithm — every enabled algorithm also appears as a filter, so you can review just the Invalid Files, or just what Audio Content Comparison found

Reviewing one algorithm at a time is the fastest way to build confidence in a new configuration.

Selecting and checking

Selection (which rows are highlighted) and checking (which rows will be processed) are separate. Select rows, then apply an action to the selection from the toolbar or the Selection menu:

• Check Selected (C) — mark for processing
• Uncheck Selected (U) — spare them
• Toggle Selected (T) — flip each
• Invert Selection (I) — select everything not currently selected
• Clear Selection (Backspace)

Rows can be selected by clicking, shift-clicking a range, or dragging a lasso. See Keyboard Shortcuts for the full set including the modifier-click variants.

Reveal Selected Files (Cmd/Ctrl+R) opens the selected files in Finder or Explorer — the quickest way to sanity-check a decision against the real file.

Save Results and Load Results in the File menu let you stop a review partway through and return to it, rather than re-running a long scan.

Removing Duplicates

Process Records applies your decisions. What actually happens depends on two independent settings — one for the database, one for the files on disk.

The database

• Create Safety Database (default) — writes a new "thinned" database without the removed records and leaves your original completely untouched. The with tag field next to it names the new file, and defaults to thinned.
• Modify Current Database — removes the records from the open database directly. Faster, and not reversible. A blinking warning icon appears next to this setting whenever it's active, because it is the one choice here that can't be undone.

The default exists for a reason: you can open the thinned database in Soundminer, confirm it's what you wanted, and still have the original if it isn't.

The files on disk

• Keep on Disk (default) — only database records change; not a single audio file is touched
• Move to Archive Folder — relocates the files to a folder you choose, so they're out of the library but recoverable
• Move to Trash — recoverable through your system Trash or Recycle Bin
• Permanently Delete — immediate and unrecoverable

Strip Dual Mono

Separate from removal. When Dual Mono Check is enabled, a Preserve Dual Mono / Strip Dual Mono selector appears above the results. Set to Strip, the flagged files are rewritten as true mono instead of being removed — you keep the sound and reclaim the space.

Afterwards, the affected records are marked dirty in Soundminer. To finish the job, open Soundminer and run: Database → Show Dirty, select all (Cmd+A), then Database → Embed Selected and Database → Rebuild Waveforms for Selected.

Tags Manager

Two editable tag lists drive the tag-based algorithms. Both live under Settings → Tags Manager.

Audiosuite Tags

The render suffixes Pro Tools adds when you process a clip — -PiSh_, -TmShft_, -NORM_, -RVRS_ and dozens more ship as defaults. The Audiosuite Tags algorithm uses this list to spot processed files and look for their unprocessed originals.

These files can be handled two ways, which you choose per tag:

• Group with Source Audio — treat the processed file as a duplicate of its original and let Preservation Priority decide
• Mark for Removal — always flag it, regardless of whether the original was found

Use Move and Copy to shift tags between the two lists, singly or all at once.

Filename Tags

Your own naming conventions, used by the Filename Contains Tag algorithm. Anything you consistently put in the names of temporary or working files belongs here.

If you use a plugin whose render tag isn't in the defaults, add it — the list is just text, and new entries take effect on the next search.

Safe Folders & Playlists

Some duplication is intentional. A sampler instrument library needs its own copies of sounds; a delivered show folder should stay exactly as delivered.

Both settings are under Settings → Match Criteria.

Safe/Ignore Folders lists paths that are excluded from duplicate searches entirely — files inside them are never flagged.

Protect Audio Files in Playlists extends the same protection to any file used in a Soundminer playlist, so building a playlist is enough to shield a file from removal.

Audio Fingerprints

Found under Settings → Match Criteria, with the same toggles mirrored in Options → Audio Fingerprints.

Audio Content Comparison works by fingerprinting each file and comparing fingerprints. Generating them is the slow part, so SMDB Companion can store them and reuse them.

• Store audio fingerprints in database — save fingerprints as they're computed
• Fetch stored audio fingerprints from database — reuse them instead of recomputing

With both enabled, the first content comparison on a library is slow and every one after it is dramatically faster. Leave them on unless you have a reason not to.

Compare Algorithm

• Exact Match — the audio must be identical. Fastest and safest.
• Relative Match — allows variation, controlled by the Threshold below. Catches re-encoded or slightly edited copies.
• Subset Match — finds files whose audio appears inside a longer file, for catching clips pulled from a longer take.

Threshold (0–100%) sets how similar is similar enough, and applies to Relative and Subset matching. The default of 80% is a reasonable starting point; raise it if you're getting false matches, lower it if you're missing obvious ones.

Clear Audio Fingerprints from Database deletes stored fingerprints, either all of them or just the selected records. This is not undoable — the fingerprints have to be regenerated — but it's harmless beyond the lost time, and useful if files have been replaced since they were scanned.

Metadata Tools

Find/Replace Metadata (Cmd/Ctrl+D) edits metadata across your database in bulk, independently of duplicate searching.

The classic use: you moved a folder of files and Soundminer's stored paths are now wrong. Rather than rescanning everything, find the old path fragment and replace it with the new one.

You specify:

• Find — the text to look for
• Replace — what to put in its place
• Column — which database field to edit; any column in your database is available
• Case Sensitive — whether case must match
• Mark Records as Dirty — flag changed records so Soundminer knows to re-embed them

Leave Mark Records as Dirty on when the change should be written back into the files themselves. Afterwards, in Soundminer: Database → Show Dirty, select all, then Embed Selected.

Presets

A preset stores your entire configuration — algorithms, match criteria, preservation rules, tags, colours. Save, load, and delete them from the bar at the bottom of Settings, or load one from Options → Load Preset.

Two ship with the app. default is the conservative configuration described throughout this guide. TJF is the developer's own working setup — considerably more aggressive, with Similar Filename, Audiosuite Tags, and Filename Contains Tag all enabled, filename-only matching, extra tags, and preservation rules built around a specific folder structure.

To start over, load the default preset.

Appearance

Colours are configurable throughout, and they carry meaning in the results table — which record is being kept, which is being removed, what needs attention. Pick a ready-made colour scheme or set individual colours under Settings → Appearance.

Font size is adjustable, and Show Toolbars (,) hides the toolbars to give the results table more room on a small display.

Keyboard Shortcuts

All shortcuts are remappable under Settings → Keyboard Shortcuts. Defaults:

Files & search

Cmd/Ctrl+O	Open Database
Cmd/Ctrl+Shift+O	Open Recent
Cmd/Ctrl+W	Close Database
Cmd/Ctrl+Enter	Search Database
Esc	Cancel Search
Cmd/Ctrl+S	Database Search window
Cmd/Ctrl+F	File System Search window
Cmd/Ctrl+D	Find/Replace Metadata
Cmd/Ctrl+Backspace	Remove Records
Cmd/Ctrl+R	Reveal Selected Files
Shift+Cmd/Ctrl+R	Reveal Database Folder
Cmd/Ctrl+,	Settings

Views

2 / 3 / 4 / 5	Results / Split / No Frills / Registration
S	Toggle search bar
N	Toggle No Frills
,	Toggle toolbars
F1	Help

Selection

C / U / T	Check / Uncheck / Toggle selected
I	Invert selection
Backspace	Clear selection
Click	Toggle row
Alt+Click	Toggle all
Shift+Click	Select range
Cmd/Ctrl+Shift+Click	Unselect range
Drag	Lasso select
Cmd/Ctrl+Drag	Lasso unselect

Tips & Best Practices

• Back up before your first real run. The defaults are non-destructive, but your settings are yours to get wrong.
• Start with Invalid Files. On any long-lived database it finds real problems immediately and can't cost you anything — those files are already gone.
• Add one algorithm at a time. Turning everything on at once produces a result set too large to review honestly, and you won't know which setting caused what.
• Spend your time on Preservation Priority. It has more effect on result quality than any other setting, because it's the part that knows about your library.
• Use the algorithm filters. Reviewing one algorithm's findings at a time is far more reliable than scrolling one long list.
• Keep the safety database on until you've confirmed a thinned database opens correctly in Soundminer.
• Leave fingerprint storage enabled. The first audio comparison is slow; every later one is fast.
• Reveal files when unsure. Cmd/Ctrl+R takes a second and settles the question.
• Not every duplicate should go. Sampler libraries and delivered shows belong in Safe Folders.

Support

Need help? Contact us:

• Email: [email protected]
• Website: feralfreq.com/smdbc
• FAQ and video overview

For license recovery, visit the license recovery page.