Best algorithms of Voice process.

The AI Sound Technology.

Let us introduce you the most advanced sound technology.
You will see the AI processing effect of all sounds.
Let your infinite imagination come into play!

The Intelligent Ear

Amazing sound process is all in here.

You hear and I can do everything you want…
– Speech Recognition
– Acoustic Event Recognition
– Direction of arrival
– Robustness in Adverse Conditions
(Ex. Audio enhancement and Blind source separation)

Play Video

Real-Time De-noise

Can noise-reduction be equated to de-noise?

In our definition, noise reduction cannot be equated to de-noise.
The lightweight algorithm can make all kinds of sounds disappear completely in real-time and leaving only specific sounds.
You can put it in SoC, this is also very suitable for real-time de-noise when using a mobile phone call or real-time de-noise when using a teleconferencing system,
Just like in the demo, when a person is chatting on the phone outside, the background sound is a lot of noisy sounds. This technology can still make the background sound completely disappear in real time.
We can even convert your speech into other people’s speech in real time.

De-noise and Anti-fake

Identify the real human speech.

Have you heard of human voice anti-counterfeiting?
We can identify where the voice comes from, such as the sound from TV, radio, and use any type of recording, 
Then there are different applications according to different scenario.
Just like in the demo, a person is chatting on the phone, and the background sound is the human voice playing news.
This technology can make the background human voice disappear completely in real time.
In addition to being very suitable for real-time de-noise using mobile phones or teleconference systems, this technology can even help voice recognition to prevent counterfeiting and recognition.

Play Video
Play Video

Speech process

Real-time human speech processing.

Why do we need to separate the human voice?
You can imagine that the human voice will become more natural without mechanical sound after some processing such as equalizer etc.

Real-time Music separation

Everyone can sing very well

When you hear a song and want to sing, you can eliminate the singer’s voice and keep only the accompaniment.
This technology can also score based on your pitch, and even automatically help you tune into the correct pitch.
Everyone can become a singing expert.

Play Video
Play Video

Auto speech recognition

We can also recognize dialects and even edge computing.

Just like the demo video,
We can process “Automatic Speech Recognition (ASR)” very accurately in real time/offline,
This is also very suitable for general voice into text records!
We can even deal with dialects,
Just like the Taiwanese dialect(TaiYu) in the second half of the movie,
our training model can be processed very accurately.

ASR & Auto-translation

This is a translation combined with localized language usage.

Praesent a nostra lacus aliquet torquent molestie class ullamcorper leo vulputate ut nascetur magnis sociis sodales condimentum mus dignissim a mi sem scelerisque ligula fermentum sociosqu et cursus parturient nostra ultricies vehicula.

Play Video

Speaker Identification

In addition to speaker recognition, then?

This technology can tell who is speaking through voiceprints.
You can use this technology with ASR to automatically generate text meeting records.
If this technology is paired with a microphone array, we can even accurately locate the speaker’s position(Direction of Arrival estimation, D.O.A) and automatically balance the volume control based on the distance between the speaker and the microphone.
Of course, this can also be used in the identification of the speaker’s voiceprint.
Don’t forget that the voice de-noise &  anti-counterfeiting technology was also introduced in the previous paragraph, and it can also be used in conjunction.

Play Video

Face identification, D.O.A and Speaker Recognition.

Play Video

The identification of the speaker’s voiceprint.

Play Video

Speaker emotion Recognition

7 classes of emotion can be recognized

In addition to using visual algorithms to see how your current emotions are,
we can also recognize your emotions through your voice.
Our technology can recognize 7 different voice emotions,
angry, boredom, disgust, fear, joy, Neutral, and sadness.
This is very suitable for giving follow-up suggestions when you are doing general conversations or meetings.

Sound recognition

It can recognize many different kinds of sounds

There are many types of sound recognition that we have processed.
It can be used in many different industries, such as factories, homes, shops, hospitals, etc.
Our technology can recognize the sound of footsteps, pulling a chair, baby crying, dog barking, broken glass, knock, and several kinds of alarm(smoke/fire/type of medical treatment and factories),etc.
As demonstrated in the video, we can also recognize the warning sounds of various medical instruments in the ward, and the abnormal noise from the machine fan.

Play Video

Sound recognition of ward for Health care.

Play Video

Sound recognition of several environments.

Play Video

Abnormal sound detection of machine fan.

Application of home health care

The Advanced Cough & Heart breath detection.

Following the sound recognition described in the previous paragraph, here we use a demonstration video about home health care.
We combine the mmWave technology, in addition to detecting human heart rhythm,
it can also automatically detect the frequency and classification of coughing, snoring and heart breathing(even the Sleep Apnea Syndrome);
This non-invasive device can automatically record the physiological condition of the subject,
provide doctors with more accurate diagnosis and treatment, and has been adopted by many hospitals and insurance companies.

Play Video

Heart breath detection for Health care.

Play Video

Cough detection for Health care.