Skip to content
Contact us
DBM10L_chip

Efficient & Powerful

Synaptics DBM10L Dual Core Edge AI SoC (DSP+NPU)
NEURAL_NETWORK

NEURAL
NETWORK

 
ARTIFICIAL_INTELLIGENCE

ARTIFICAIL INTELLIGENCE

 
ULTRA-LOW_POWER

ULTRA-LOW POWER

 
machine_learning

MACHINE
LEARNING

 
DBM10L_chip

Efficient & Powerful

Synaptics DBM10L Dual Core Edge AI SoC (DSP+NPU)

NEURAL
NETWORK

ARTIFICIAL
INTELLIGENCE

ULTRA-LOW
POWER

MACHINE
LEARNING

Optimized for Voice &
Sensor Processing

The DBM10L enables artificial intelligence (AI) and
machine learning (ML) based sensor data functions

VOICE_TRIGGER
VOICE_AUTHENTICATION
VOICE_COMMAND

VOICE
TRIGGER

VOICE
AUTHENTICATION

VOICE
COMMAND

ACOUSTIC_ECHO
SOUND_EVENT
NOISE_REDUCTION

ACOUSTIC ECHO
CANCELLATION

SOUND EVENT
DETECTION

NOISE
REDUCTION

Optimized for Voice &
Sensor Processing
 
The DBM10L enables artificial intelligence (AI) and
machine learning (ML) based sensor data functions
 

VOICE
TRIGGER

VOICE
AUTHENTICATION

VOICE
COMMAND

ACOUSTIC ECHO
CANCELLATION

SOUND EVENT
DETECTION

NOISE
REDUCTION

smart-home

Extended Battery Life for High Performance Devices

checkmark-1 Tablets / Handheld

checkmark-1Security Sensors

checkmark-1 Smart Home

checkmark-1 Wearables

checkmark-1 Remote Controls

checkmark
Tablets / Handheld
checkmark
Security Sensors
checkmark
Smart Home
checkmark
Wearables
checkmark
Remote Controls

Start Your Designs Today...

DBM10L_chip-1

+ Toolchain + EVK

DBM10L_chip
+ Toolchain + EVK

The Synaptics DBM10L is a dual-core SoC comprising a DSP and a neural processing unit (NPU). It is available today to designers and partners, complete with the entire toolchain, including our proprietary nNet Lite™ neural network (NN) processor, a dedicated hardware engine, designed to accelerate inference at ultra-low power consumption.

nNet Lite - Ultra-Low-Power Neural Network Inference Processor

Multi-Input / Sensor NNs

Able to process inputs and run inference on data from multiple sensors such as those for speech, sound, vibration, and temperature 

Easy Porting / Easy to Use

Full support of all standard DNN frameworks, plus comprehensive, cross-platform toolchain for model migration and optimization

Efficient Model Size Optimization

Allows porting of large models, without loss of accuracy, utilizes 1~16 bit quantization, post-training model pruning, and lossless entropy compression algorithms

Multi-Input / Sensor NNs

Able to process inputs and run
inference on data from multiple
sensors such as those for speech,
sound, vibration, and temperature 

Easy Porting / Easy to Use

Full support of all standard DNN
frameworks, plus comprehensive,
cross-platform toolchain for model
migration and optimization

Efficient Model Size Optimization

Allows porting of large models, without
loss of accuracy, utilizes 1~16 bit
quantization, post-training model
pruning, and lossless entropy
compression algorithms

nNet Lite - Compatible With All Standard DNN Frameworks
chainer_red_h
mxnet_logo_2
og_image_caffe2
Microsoft-CogTK
tensorflow_logo
Keras_logo
Gluon_Logo
PyTorch_logo
Theano_logo

Contact us to learn more about our DBM10L