YAMS
Yet Another MotionSenSE Service utility — run collections on the wristband, pull data off it, and turn raw sensor binaries into CSV.
One app for the whole capture workflow
YAMS opens in your browser as a set of tabs. Each one covers a stage of working with the MotionSenSE wristband, from starting a collection to getting analysis-ready CSVs.
Run a collection
Scan for wristbands, select one or more to connect, then start and stop recording while watching collection state live.
Pull data off the device
Browse onboard sessions and download them, optionally extracting to CSV in the same step.
Binaries to CSV
Point it at a folder of .bin files and get one CSV per sensor per subject.
Zip in, zip out
Drop in a zip of device folders and get an extracted archive back, without unpacking by hand.
Check it before you trust it
Plot extracted signals to confirm a capture looks right before it reaches your analysis.
Align the clocks
Reconcile device time against reference timestamps so streams line up across devices.
Read device identity
Pull serial numbers, firmware versions and identifiers out of connected wristbands.
Devices & sessions
Manage the device registry and the subject/session encoding table used to name outputs.
Three ways to get it
If you just want to use YAMS, take the first one — it needs no Python at all.
Prebuilt app
No Python installation needed. On Windows, use the installer or portable zip.
YAMS_Windows_x64_Setup.exe for Windows, YAMS_MacOS_arm64 for
Apple silicon Macs. Intel Macs are not covered by the prebuilt
build — use pip or source instead.
From PyPI
If you already work in Python.
pip install -U yams-util python -m yams
From source
For development, or to run unreleased changes.
git clone https://github.com/SenSE-Lab-OSU/YAMS.git cd YAMS git checkout v2 pip install -e . conda install -c conda-forge liblsl python -m yams
liblsl.
The Lab Streaming Layer library is a native dependency that pip cannot supply.
Installing it from conda-forge is the step most source installs miss, and without
it the app fails at startup.
However you install it, YAMS starts a local server and opens your browser at
http://127.0.0.1:7860. Nothing leaves your machine.
From wristband to CSV
The normal arc of a study session, in four steps.
-
Collect
In ⌚️ MotionSenSE controller, scan, select the wristbands to connect, and start the collection. Data is written to onboard flash, so the devices do not need to stay in range once recording has begun.
-
Download
In 📂 File downloader, pick the sessions you want and pull them off the device. Tick Extract data after download to get CSVs in the same pass. See Extract onboard data for the full walkthrough.
-
Extract
Or extract separately in 🛠️ Data extractor: give it an input folder of
.binfiles and an output folder. Record formats are detected automatically, so there is normally nothing to configure. -
Verify
Open the result in 📊 Data viewer, and read
README.txtin the output folder — it records how every single file was decoded.
Formats are detected, not assumed
The wristband has written six different on-disk record layouts over its life, and the firmware version recorded on the device is not always right about which one you have. YAMS reads the layout out of the file itself.
uuid.txt is stale or missing entirely now extract correctly with
no flags at all — previously they produced silently wrong numbers.
Run it from the command line:
python -m yams.data_extraction -i path/to/binaries -o path/to/output
Or inspect what would happen first, without writing anything:
python -m yams.data_extraction -i path/to/binaries --dry_run
The options you are most likely to need:
| Flag | Default | What it does |
|---|---|---|
--dry_run | off | Report the layout resolved for every binary and exit |
--validate_with_uuid | off | Cross-check detection against uuid.txt and report disagreements |
--ppg_format | auto | Force a PPG layout, or version for the pre-1.6 behaviour |
--save_format | csv | csv or pickle |
--mode | dir | dir for a folder of binaries, batch for a folder of zips |
Every flag, all six record layouts, how detection works and how it is measured: Data Extraction reference →