General Information
These release notes contain important information about CCStudio™ Edge AI Studio v1.8.0 that may not be included in other product documentation. Please review this document before using this release.
Edge AI Studio v1.8.0 is a feature release. It adds new audio and Wi-Fi sensing applications (Wi-Fi CSI Presence Detection, Audio Event Detection, Cough Detection, Google Speech Commands, Wake Word Detection, Glass Break Detection), a Dataset Format Checker, a new Feature Extraction Pipeline GUI, model quantization options, training comparison, parallel training runs, and includes various enhancements and bug fixes. For a full summary of changes, see the What’s New section below.
System Requirements
The Edge AI Studio desktop installer is available for 64-bit Windows and 64-bit Linux. The cloud version is supported on any modern browser (Chrome recommended) running on Windows, Linux, or macOS.
- Windows 11 64-bit
- Windows 10 64-bit
- 8 GB RAM minimum (16 GB recommended)
- 10 GB free disk space
- 64-bit Linux (Ubuntu 20.04 LTS or later recommended)
- Python 3 with pip pre-installed
- 8 GB RAM minimum (16 GB recommended)
- 10 GB free disk space
- Google Chrome 120+ (recommended)
- Microsoft Edge 120+
- Firefox 120+
- Internet connection required
Hardware Requirements for Data Capture
To use the Sensor BoosterPack™ live capture and live preview features, a compatible TI LaunchPad™ and the BOOSTXL-SENSORS BoosterPack are required:
- Compatible LaunchPad™ connected via USB
- BOOSTXL-SENSORS BoosterPack™ (provides accelerometer, gyroscope, microphone, PIR, ambient light, humidity, and temperature sensors)
Installation
Windows Desktop Installer
Download the installer from the link below and run it to install Edge AI Studio on your Windows machine.
- Run
eaistudio-1.8.0-windows-x64-installer.exeas Administrator. - Accept the license agreement and choose an installation directory.
- Wait for all components to install, then click Finish.
- Launch Edge AI Studio from the Start Menu or desktop shortcut.
The SDK dependency checker will not reflect the v1.8.0 SDK until the corresponding SDK is available on Resource Explorer. In the meantime, you can manually install the required SDK version. To update the dependency checker, edit the rex_dependencies section in <installDir>\model-composer-extensions\tinyml-backend-1.5.0\overrides.json.
Linux Desktop Installer
Download the installer from the link below and run it to install Edge AI Studio on your Linux machine.
- Make the installer executable:
chmod +x eaistudio-1.8.0-linux-x64-installer.run - Run
./eaistudio-1.8.0-linux-x64-installer.run. - Accept the license agreement and choose an installation directory (defaults to your home directory).
- Launch Edge AI Studio from the installation directory or desktop shortcut.
The Linux installer requires Python 3 with pip already installed on the system. Required Python packages (pandas, numpy) are installed automatically at first run rather than bundled, as they are on Windows.
Cloud Version
No installation is required for the cloud version. Navigate to the link below and sign in with your TI account:
The cloud version provides full data capture, annotation, training, and deployment workflow support directly in your browser.
Edge AI Studio v1.8.0 — What’s New
Enhancements & Highlights
- New Applications
- Wi-Fi CSI Presence Detection (learn more)
- Audio Event Detection
- Cough Detection
- Google Speech Commands
- Wake Word Detection
- Glass Break Detection
- Visualization for Wi-Fi Channel State Information (CSI) data format
- Built-in playback for WAV file format
- Dataset Format Checker: reports errors and warnings for datafiles in an incorrect format
- New Feature Extraction Pipeline GUI for new Training Runs
- Model Quantization: 2, 4, or 8 bits, or Mixed Auto Precision
- Training Comparison: select two existing training runs to compare
- Parallel Training Runs: run up to 4 training runs in parallel
- Various enhancements and bug fixes
Bug Fixes
Supported Devices
Microcontroller
| Family | Devices | GPNs |
|---|---|---|
| AM series | AM13E2, AM261, AM263, AM263P | AM13E23019, AM2612, AM2612-Q1, AM2631, AM2631-Q1, AM2632, AM2632-Q1, AM2634, AM2634-Q1, AM263P2, AM263P4, AM263P4-Q1 |
| CC series | CC1352, CC1354, CC2745, CC2755, CC35X1 | CC1352R, CC1352P, CC1352P7, CC1354R10, CC1354P10, CC2745R10, CC2755R10, CC3501E, CC3551E |
| F28 series | F280013, F280015, F28003, F28004, F2837, F28P55, F28P65 | TMS320F2800133, TMS320F2800135, TMS320F2800137, TMS320F2800155, TMS320F2800157, TMS320F280033, TMS320F280034, TMS320F280037, TMS320F280037C, TMS320F280039C, TMS320F280039, TMS320F280041, TMS320F280041C, TMS320F280045, TMS320F280049C, TMS320F28374S, TMS320F28374D, TMS320F28375S, TMS320F28375D, TMS320F28376S, TMS320F28376D, TMS320F28377S, TMS320F28377D, TMS320F28378D, TMS320F28379S, TMS320F28379D, TMS320F28P550SJ, TMS320F28P650DH, TMS320F28P650SK, TMS320F28P650DK |
| F29 series | F29H85, F29P32, F29P58 | F29H850TU, F29P329SM-Q1, F29H859TU-Q1 |
| MSPM0 series | MSPM0G3507, MSPM0G3519, MSPM0G5187 | MSPM0G3507, MSPM0G3519, MSPM0G5187 |
| MSPM33 series | MSPM33C32 | MSPM33C321A |
Recommended SDKs & Compilers
The following SDKs and compiler toolchains are recommended to compile and deploy models generated by Edge AI Studio v1.8.0. Install the listed version (or a newer compatible version) for your target device family.
SDKs & Compilers
| SDK / Compiler | Version | Supported Devices |
|---|---|---|
| AM13E230X SDK | 26.01.00.03 | AM13E2 |
| SimpleLink SDK EdgeAI Plugin | 1.20.00.00 | CC1312, CC1314, CC1352, CC1354, CC2745, CC2755, CC35X1 |
| SimpleLink CC13xx CC26xx SDK | 8.33.00.16 | CC1312, CC1314, CC1352, CC1354 |
| SimpleLink Low Power F3 SDK | 9.20.00.81 | CC2745, CC2755 |
| SimpleLink™ Wi-Fi SDK | 10.20.00.39 | CC35X1 |
| MCU+ SDK for AM263x | 26.00.00.06 | AM263 |
| MCU+ SDK for AM263Px | 26.00.00.06 | AM263P |
| MCU+ SDK for AM261x | 26.00.00.06 | AM261 |
| MSPM0 SDK | 2.11.00.07 | MSPM0G3507, MSPM0G3519, MSPM0G5187 |
| C2000Ware | 26.01.00.00 | F280013, F280015, F28003, F28004, F2837, F28P65, F28P55 |
| C2000Ware MotorControl SDK | 5.04.00.00 | F280013, F280015, F28003, F28004, F2837, F28P65, F28P55 |
| C2000Ware DigitalPower SDK | 26.01.00.00 | F280013, F280015, F28003, F28004, F2837, F28P65, F28P55 |
| F29H85X SDK | 26.00.00.00 | F29H85, F29P58, F29P32 |
| MSPM33 SDK | 1.03.00.01 | MSPM33C32, MSPM33C34 |
| TI C28-CLA Compiler | 25.11.01.00 | F280013, F280015, F28003, F28004, F2837, F28P65, F28P55 |
| TI C29 Clang Compiler | 2.02.00.00 | F29H85, F29P58, F29P32 |
| TI Arm Clang Compiler | 5.01.01.00 | AM13E2, AM263, AM263P, AM261, CC1312, CC1314, CC1352, CC1354, CC2745, CC2755, CC35X1, MSPM0G3507, MSPM0G3519, MSPM0G5187, MSPM33C32, MSPM33C34 |
Previous Releases
The table below summarizes previous Edge AI Studio releases with their key highlights.
View previous release notes
| Version | Type | Highlights |
|---|---|---|
| v1.7.1 | Maintenance | CC2745 device support; Classification, Regression, Forecasting, and Anomaly Detection support extended to CC1312, CC1314, CC1352, CC1354, CC2755, and CC35xx; new Getting Started example projects documentation (MOSFET Temperature Prediction and Grid Fault Detection) |
| v1.7.0 | Feature | Anomaly Detection task support; new dataset and training analysis reports (Goodness of Fit, One vs Rest Multiclass ROC, Histogram Class Score Differences, PCE on Feature Extracted Clusters, Distribution of Reconstructed Error); new example projects; Linux desktop platform support |
| v1.6.1 | Maintenance | Cloud availability for Time Series workflows; Device Detection support for Capture and Training; Confusion Matrix for Training analysis; fixes for F28 and F29 training runs |
| v1.6.0 | Feature | Additional Time-series AI tasks (Regression, Forecasting); Sensor BoosterPack™ support for live data capture; expanded MCU device coverage (F29, AM13, AM26, MSPM0, MSPM33, CC13xx/26xx, CC23xx, CC27xx, CC35xx); example projects library; revamped workspace and training workflows |
| v1.5.0 | Feature | mmWave Point-cloud Classification with Pose Estimation; xWRL6432 Radar device support; MSPM0, MSPM33 device support; cloud deployment enhancements |
| v1.4.0 | Feature | Time-series anomaly detection; arc-fault and motor-bearing fault detection; desktop (Windows) support; expanded F28 device family coverage |
| v1.3.0 | Feature | Cloud-only time-series support; initial MCU device coverage; AM, F28 series support; BYOD data capture workflow |
| v1.2.0 | Feature | Vision AI model training; AM62, AM62A, TDA4VM device support; image annotation tools; model compilation and deployment workflow |
All Open Issues
For a complete list of currently unresolved issues in Edge AI Studio, use the link below.