The following are the features and engine support status provided by ThinQ.AI.
Speech & Language
Hybrid Trigger Word Detection
Detects a specific trigger word in the user's voice input from the microphone and further checks the result on the server.
Automatic Speech Recognition
Processes and converts the user's voice input from the microphone into text and delivers the result.
Natural Language Processing
Converts unstructured natural language into structured language data that the machine can understand, and conveys the required information for the engine operation and the user's intention.
Text to Speech
Converts text data into a speech and plays the generated PCM file.
Speech Signal Processing
Removes the noise from the voice signal converted to digital. It also executes one- or two-channel preprocessing to process voice signals.
Multi-device Wakeup Arbitration
If multiple devices recognize the same trigger word at the same time, this engine selects the best device for the user.
Recognizes the user's emotions through the input data (voice, image, and text) and returns the user's emotional state.
Detects whether faces exist in an input image and returns the result.
Detects faces from an input image and registers/removes the face to/from the album database.
Detects hands from an input image and returns the hand position and gesture.
Age Gender Recognition
Detects faces from an input image and returns the estimated values of age and gender.
Driver State Monitoring
Detects the facial area, including eyes from input data, calculates the landmarks of both eyes, and returns the state information of the driver's eyes.
Allows combination of multiple authentication methods for robust security along with convenience.
Detects snoring and apnea during sleep and analyzes the patterns and phases of sleep.
Cloud Deep Learning Toolkit
This is a web-based program that allows developers to train AI engines using a dataset suitable for their own purpose.
Feedback Analysis System
This is a tool that allows developers to analyze user feedback on learned models.
Embedded Deep Neural Network
This is a Model Compression program that reduces the model size so that the trained model can run in the embedded environment.
Platform Development Kit
This is a development kit that automatically generates the necessary codes to make ThinQ.AI SDK development easier.
Open Innovation Technology
Open Innovation Technology
This is a service that combines the technologies of ThinQ.AI with those of partners. ThinQ.AI helps developers produce new products using OI technologies.
AI Chip: LG8111
LG8111 AI Chip supports artificial intelligence functions such as visual/voice/control intelligence. It is equipped with LG Neural Engine (LNE) which is an AI processor that mimics the neural network of the human brain.
This is the specification of the ThinQ.AI SDK and Cloud API.
The hardware support and performance required for each engine varies by OS. Therefore, there may be a difference in engines and performance supported by OS.
|Speech & Language||Hybrid Trigger Word Detection (HTWD)||○||○||○||○||○|
|Automatic Speech Recognition (ASR)||○||○||○||○||○|
|Natural Language Processing (NLP)||○||○||○||○||○|
|Text to Speech (TTS)||○||○||○||○||○|
|Speech Signal Processing (SSP)||○||○||○||X||X|
|Multi-device Wakeup Arbitration (MWA)||○||○||○||X||X|
|Emotion||Emotion Recognition (ER)||○||○||X||X||○|
|Vision||Face Detection (FD)||○||○||○||X||X|
|Face Recognition (FR)||○||○||○||X||X|
|Hand Gesture (HG)||○||○||○||X||X|
|Age Gender Recognition (AGR)||○||○||○||X||X|
|Driver State Monitoring (DSM)||○||X||X||X||X|
|Biometrics||Multi-modal Biometrics (MMB)||○||○||X||X||X|
|Sleep Detection (SD)||○||○||X||X||X|