“Ok! Google! Play the music!” It is close to such a day when I hear it in Japan too. Google revealed that Google Home will be available in the Japan market in this summer of 2017.
Using the Google Home product, by voice control you can play music, send email, check traffic conditions and even control lighting in a room away from each other in conjunction with a Google home device in each room.
The other day, as reported on news that NXP focuses on it and has partnered with Amazon. NXP seems to be very active in a voice recognition technology.
” LONDON–NXP Semiconductors has partnered with Amazon to make Alexa voice-recognition hardware and software available as a reference design for integration into other devices, part of a wave of chip makers looking to cash in on the success of digital assistants in hopes that OEMs will begin incorporating voice recognition and command capabilities in other products.”
EETimes, NXP Reference Design Integrates Alexa Voice Control Into Appliances, 2017/April 17
This news seems be based on i.MX, but NXP is focusing on the technology not only for i.MX processors but also for microcontrollers. Voice recognition is also supported in the LPC5411x series.
Let’s look in LPC 5411x series this time. Well, let’s go from the points.
- LPC5411x is Dual-core architecture
- DMIC(Digital microphone subsystem) is supported to realize voice recognition.
- Power efficiency is very high.
LPC5411x Block diagram
Source: NXP LPC5411x Product Page
Block diagram gives us a summary of the device and what it is like.
So it is important.
The greatest feature of the LPC 541 xx series is the dual core architecture of the ARM Cortex-M0+/Cortex-M4F, but it consumes ultra-low power consumption, achieving 80 μA/MHz for Cortex-M4F and 60 uA/MHz for Cortex-M0+.
It is surprisingly low power consumption!
Until a few years ago, considering that it was around 200uA/MHz, the efficiency has been dramatically improved to as half as those days.
Digital Microphone Subsystem(DMIC)
In addition, it has a Digital Microphone Subsystem(Option supported by LPC5411x), and it is an interface for digital microphone, and that can capture a sound using a digital microphone and can recognize a voice signal.
The voice recognition engine will be processed by software, and it is provided by 3rd party, Sensory.
Best use-case is a sensor application with small form factor.
With the asymmetric dual core architecture (Cortex-M4F/Cortex-M0+) which can be said as a heterogeneous, Cortex-M0+ periodically monitors sensor data, while Cortex-M4F performs an algorithm processing of captured sensor data. That is, the cores are divided in right roles for processing data.
By doing this, an ultra low power consumption can be realized by performing appropriate processing in each core and by dividing the right roles.
It gives the best use-case in a small form factor of sensor applications such as mobile, wearable, fitness and healthcare.
Another feature is a voice recognition technology, that is, a voice can be detected very first without running the core. In other words, LPC5411x can detect a voice while a core remaining in sleep.
For voice detection, HWVAD (Hardware Voice Activity Detection) and various filters are built in by hardware, and it is a technology to detect a voice by using a voice envelope, noise floor, etc.
When detected as the first voice, the interrupt is generated to wake up the CPU core and it is activated to process. Subsequently, it is a flow to perform voice recognition and voice trigger control by CPU core processing voice algorithm engine.
Currently, NXP promotes using Sensory’s TrulyHandsFree voice trigger engine.
The LPCXpresso board, OM13090 makes it possible to evaluate this voice recognition technology.
In another article, I would like to introduce this OM 13090 and will try to get it up and running.
NXP seems to pitch in a voice recognition technology, and there is such a news that collaborates with Amazon in the area of a voice recognition. Even in the microcontroller area, NXP seems to do so on LPC 541×0 series.
In the near future, I would like to write an article about the voice recognition.
Thank you so much for your reading to the end.
If you think it was good! It would be greatly appreciated if you could share it.