Ambiq ModelZoo
Models are the heart of AI - they're the bit with the intelligence in it.
AI model development follows a lifecycle - first, the data that will be used to train the model must be collected and prepared, second, the model is 'trained' on that data, and finally the trained model is compressed and deployed to the endpoint devices where they'll be put to work. Each one of these phases requires significant development and engineering.
Ambiq's ModelZoo is a collection of open source endpoint AI models packaged with all the tools needed to develop the model from scratch. It is designed to be a launching point for creating customized, production-quality models fine tuned to your needs.
Each model's repository includes:
- Data preparation scripts which help you collect the data you need, put it into the right shape, and perform any feature extraction or other pre-processing needed before it is used to train the model.
- Training scripts that specify the model architecture, train the model, and in some cases, perform training-aware model compression such as quantization and pruning
- Inference scripts to test the resulting model and conversion scripts that export it into something that can be deployed on Ambiq's hardware platforms.
- Where possible, pre-trained deployable models, and power-optimized firmware connecting to Apollo4's sensors and peripherals.
The Models
Built on top of neuralSPOT, our models take advantage of the Apollo4 family's amazing power efficiency to accomplish common, practical endpoint AI tasks such as speech processing and health monitoring.
| Model | Description | Sensor | Pre-trained? | Real-Time? |
|---|---|---|---|---|
| NNSP | Neural Network Speech Processing - 3-in-1 model including voice activity detection, keyword spotting, and speech-to-intent | Analog or digital microphone | no | Yes |
| NNSE | Neural Network Speech Enhancer - removes background noise from speech. | Analog or digital microphone | Yes | Yes |
| NNID | Neural Network Speaker Identification - identifies a speaker based on their speech pattern. | Analog or digital microphone | Yes | Yes |
| ESC | Environmental Sound Classification - identifies sounds such as crying babies, barking dogs, and gunshots. | Analog or digital microphone | Yes | Yes |
| HeartKit | Analyzes ECG data to classify beats, identify ECG segments, and detect irregular heartbeats. | ECG over I2C | Yes | Yes |
| HAR | Human Activity Recognition - detects activities such as running, walking, sitting, and climbing stairs. | Accelerometer + Gyroscope over I2C | Yes | Yes |
Neural Network Speech
This real-time model is actually a collection of 3 separate models that work together to implement a speech-based user interface. The Voice Activity Detector is small, efficient model that listens for speech, and ignores everything else. When it detects speech, it 'wakes up' the keyword spotter that listens for a specific keyphrase that tells the devices that it is being addressed. If the keyword is spotted, the rest of the phrase is decoded by the speech-to-intent. model, which infers the intent of the user.
Heartkit, with Heart Arrhythmia Classification
This real-time model analyzes the signal from a single-lead ECG sensor to classify beats and detect irregular heartbeats ('AFIB arrhythmia'). The model is designed to be able to detect other types of anomalies such as atrial flutter, and will be continuously extended and improved.
NOTE This is demonstration model only, not intended for medical applications.
Speech De-Noise
This real-time model processes audio containing speech, and removes non-speech noise to better isolate the main speaker's voice. Our approach is a novel combination of speech enhancement techniques that result in a very compact model producing high quality speech.
Human Activity Recognition
This real-time model analyses accelerometer and gyroscopic data to recognize a person's movement and classify it into a few types of activity such as 'walking', 'running', 'climbing stairs', etc.
Speaker Identification
This real-time model listens to speech and identifies the speaker as one of up to five speakers. New speakers are added post-training, avoiding the necessity of a full re-training of the model.
Environmental Sound Classification
This real-time model listens to sounds and identifies bark dogs, gun shots, sirens, babies crying, or 'other'. This type of model is useful for security and accessability applications.
A Word About Datasets
Our models are trained using publicly available datasets, each having different licensing constraints and requirements. Many of these datasets are free (or very cheap) to use for non-commercial purposes such as development and research, but restrict commercial use. Since trained models are at least partially derived from the dataset, these restrictions apply to them.
Where possible, our ModelZoo include the pre-trained model. If dataset licenses prevent that, the scripts and documentation walk through the process of acquiring the dataset and training the model.