Last Updated: Jan 10, 2026
Repository Note (Feb 22, 2026): Git history was reset after a full architecture upgrade.
Please treat the latestmaincommits as the canonical source of truth.
Welcome. To download user executable file, please visit this link.
For Chinese users: 中文用户指南与软件的下载，请参见此链接
This document contains a quick start guide for you. The picture references are based on the MacOS version of LMA.
However, Windows version of LMA follows the exact same logic, except they look a bit different.
Thank you for using LMA (Working Title) Local Audio Sample Manager. I designed this light tool to help music creators retrieve and tag their local audio samples faster.
- Press "Select Folder" and choose the local sample folder and MIDI folder you wish to scan.
- Notice that LMA require you to put your target files into one folder for each category.
- Press "Rescan" to perform a fresh scan of the samples or MIDIs within the folder.
- When the amount of files are large, it may took a while for the scan to complete. Please wait until the process finish.
- Enter your search keywords in the Enter Keywords field, then click Search or press Enter to find samples.
- Click on a sample that appears in the window below to preview the audio. Clicking it again will stop the playback.
- MIDI preview in audio is not available.
- Once a sample is selected, you can drag and drop it to your DAW or another location. The dragged sample will be copied to the new location.
- Original sample being dragged will remain on its original location.
- You can choose to search for loop or one-shot samples.
- Press the bullet list will perform a search with updated criteria immediately.
- You can search for samples within a specified BPM range.
- If you enter the same number for both the start and end of the range, the search will switch to looking for an exact BPM match.
- You can search based on key of sample. But please be aware of the following rules:
- Please note that LMA's search is based on file naming, not audio file analysis.
- Consequently, files with incorrect naming will not be actively identified, and users should be aware of this.
- Due to the complexity of naming and limitations, users must follow the syntax I have designed to conduct precise sample searches.
- When searching for a major key, please search directly for the target note name (e.g., A).
- The search results will include the major key, but may also contain other content. This will be explained later.
- When searching for a minor key, such as A minor, please use this format: Am.
- Inputting "Amin" or "Aminor" will cause search issues. This is because I have already standardized all equivalent expressions to ensure results point to the correct target.
- Searches for other modes (e.g., Lydian, Dorian) can be performed in the Keywords section, or by searching for the major key equivalent and then browsing the results.
- When searching for a major key, please search directly for the target note name (e.g., A).
- Right-click on a sample to bring up the option: "Add to Collection."
- Once selected, the sample will have a star icon next to it in the list.
- Please note that tagging a sample does not change the file's original name, so you can use this feature worry-free.
- Sample classification and indexing are based on the file's original name.
- About button will contain the current version information as well as the link to visit GitHub page of LMA.
As of Jan 10, 2026, LMA experienced an update. The update includes many changes, they will be documented below.
- 2026-02-22: Unified key parsing/normalization for sample and MIDI indexing (
B,Bb,Am,Amin,Aminorhandling). - 2026-02-22: Changed key filter to canonical exact-match search to avoid flat/natural false positives.
- 2026-02-22: Rescan now performs full index replacement to remove stale entries while preserving favorites for unchanged files.
- 2026-02-22: Switched runtime data path to
~/Library/Application Support/LMA/with legacy data migration. - 2026-02-22: Added SQLite tuning/indexes and key normalization tests.
LMA has been rewritten for performance and native compatibility. It is now built with Python 3 (Tested on 3.9+) and uses the PyQt6 framework. Core Tech Stack:
- PyQt6: The modern standard for the GUI. It handles the windowing, event loop, and native integration.
- SQLite3: Replaced the old JSON indexing system. LMA now handles libraries with tens of thousands of samples with zero lag using a local SQL database.
- QtMultimedia (PyQt6): Replaced pygame. Audio playback is now handled natively by the OS via Qt, reducing dependency issues and size.
- Threading (QThread): File scanning and database operations run on background threads to keep the UI smooth.
- soundfile: Used to parse the duration of .wav files efficiently.
⠀Code Layout:
- lma_main.py: The main entry point. It initializes the QApplication, sets up the SQLite connection, manages the UI tabs, and handles the audio player logic.
- sample_parser.py: The parsing logic for audio samples (.wav). It extracts metadata (BPM, Key, Type) from filenames and calculates duration.
- midi_parser.py: The parsing logic for MIDI files (.mid). It extracts metadata using Regex from filenames.
⠀Data Location: To comply with macOS/Windows standards and support read-only distribution (like DMGs), LMA now stores its database (lma_library.sqlite) and config (lma_config_v2.json) in the user's data directory:
- macOS: ~/Library/Application Support/LMA/
- Windows: %APPDATA%\LMA\
⠀Please let me know if you spot any bugs or have issues contributing to the code. Have fun!