SeM-V1 · ALWAYS-ON VOICE AGENT SOC
SeM-V1
Always-On Voice Agent SoC
On-Device Voice AI SoC
for Always-On Processing
SeM-V1 · ALWAYS-ON VOICE AGENT SOC
Always-On Voice Agent SoC
On-Device Voice AI SoC
for Always-On Processing
CLOUD-STREAMED VOICE PIPELINE
ON-DEVICE VOICE PIPELINE
ARCHITECTURE · ON-DEVICE VOICE INFERENCE
SeM-V1 is designed to execute KWS, wake-word detection, and STT workloads on device, so supported functions do not require continuous raw-audio streaming to the cloud.
PRODUCT DETAIL · VUI EDGE AI
Always-On Voice Agent SoC
A low-power on-device Voice AI SoC development system for KWS, wake-word detection, and STT workloads, including compiler, runtime, profiling, and evaluation hardware.
Profile operators, tensor shapes, data reuse, and latency
Partition graphs; map dataflow and memory
Couple compute clusters with local SRAM
Samsung Foundry 28 nm physical implementation
Execution path designed for continuous keyword and wake-word detection
Clustered compute and local SRAM reduce data movement
KWS, wake word, and STT run on the local device path
PRODUCT LINES
SEM PRODUCT · IN DEVELOPMENT
The image shows the SeM-V1 package concept and evaluation platforms. It is not a photograph of finished SeM-V1 silicon.
PRODUCT 02 · SeM-V2

PRODUCT 03 · SeM-V3

PRODUCT 04 · SeM-N1

SUAI LINE · TEST SILICON
A PIM-like NPU test chip built to move Semunite’s architecture into physical hardware validation.
SUAI is a separate test-chip line. It is not SeM-V1 and does not represent final SeM-V1 commercial specifications.
Assess SeM-V1 mapping feasibility from operator mix, tensor shapes, memory footprint, and latency targets.
Request Model EvaluationDefine the NPU IP, silicon, and deployment-software scope from target-product interfaces and power and area budgets.
Discuss System Integration