Posts Tagged ‘artificial intelligence’

Onkyo Has Developed a Shaker Turning Walls & Home Appliances Into Speakers

September 22nd, 2017 4 comments

According to an article on Nikkei Technology, Onkyo has designed a shaker that can be attached directly to walls and/or the surface of home appliances, and turn them into speakers to play music and voices.

The device developed using Onkyo voice & music playing technologies, has a a minimum resonance frequency of 115 hertz, and measures just 41x 36 x 21 mm. The company expects it to be used in places where visible speakers are problematic, either due to installation issues because of requirements such as waterproofness or air tightness, or for aesthetic reasons.

Onkyo will provide the shaker to OEM’s in order to be integrated into home appliances or walls for artificial intelligence, or Internet of things applications. So voice enabled walls and washing machines may just be around the corner…

Bose QuietComfort 35 II Headphones Are Optimized for Google Assistant

September 21st, 2017 3 comments

Many years ago, when the first Bluetooth headset started to be sold, it was always a bit funny to hear people apparently talking to themselves while walking in the street. But soon enough, fellow walking zombies may start talking even more on the go, as Google and its partners have launched a new category of headphones with support for Google Assistant starting with Bose QuietComfort 35 II (aka QC35 II).

The new headphones are actually an update to QC35 headphones that not only adds Google Assistant support but also noise control settings. It’s not really as cool as it first sounds though. The headphones have an “Action” button to trigger Google Assistant, so no “OK Google” or “Hey Google”, and support is not exactly built-in, meaning you’ll still need to pair the headphones with your smartphone over Bluetooth.

You’ll then be able to hear your incoming messages and calendar automatically right from your headphones, even when you listen to music. It will also be easier to control your music, for example to access specific playlists, and get informed by asking your headphones to “play the news” from your favorite news source. Finally, you’ll be able to make calls without getting your smartphone out, press the Google Assistant button, and say “Call mum” and you’re good to go.

QC35 II can also be plugged to a 1.2m audio cable, offers up to 20 hours of battery life. The headphones are charged via a USB cable, and a 15-minute charge should be good enough for around 2h30 of listening time.

Google Assistant will be available to QC35 II headphones in the US, Australia, Canada, France, Germany, and the UK for $350 or equivalent.

Via Liliputing

Imagination Announces PowerVR Series2NX Neural Network Accelerator (NNA), and PowerVR Series9XE and 9XM GPUs

September 21st, 2017 2 comments

Imagination Technologies has just made two announcements: one for their PowerVR Series2NX neural network accelerator, and the other for the new high-end GPU families: PowerVR Series9XE and 9XM.

PowerVR Series2NX neural network accelerator

Click to Enlarge

The companies claims 2NX can deliver twice the performance and half the bandwidth of nearest competitor, and it’s the first dedicated hardware solution with flexible bit-depth support from 16-bit down to 4-bit.

Key benefits of their solution (based on market data available in August 2017 from a variety of sources) include:

  • Highest inference/mW IP cores to deliver the lowest power consumption
  • Highest inference/mm2 IP cores to enable the most cost-effective solutions
  • Lowest bandwidth solution with support for fully flexible bit depth for weights and data including low bandwidth modes down to 4-bit
  • 2048 MACs/cycle in a single core, with the ability to go to higher levels with multi core

The PowerVR 2NX NNA is expected to be found in smartphone and other mobile devices leveraging Tensorflow Lite and API for Android, as well as Caffe2Go framework, smart surveillance cameras, assisted and autonomous driving solutions, and home entertainment with TVs and set-top boxes using artificial intelligence to adapt preferences to certain users. NNA will find their ways in more and more SoC as shown in the diagram below by Imagination showing the evolution of SoCs over the years, and this has already started as we’ve seen with Huawei Kirin 970 mobile SoC featuring its own neural processing unit (likely not 2NX though).

Click to Enlarge

PowerVR 2NX development resources include mapping and tuning tools, sample networks, evaluation tools and documentation leveraging industry standard machine learning frameworks such as Caffe and Tensorflow. The Imagination DNN (Deep Neural Network) API, working across multiple SoC configuration, should ease transition between CPU, GPU and NNA.

PowerVR 2NX NNA is available for licensing now which should mean products with the solution possibly coming sometimes in 2018. Some more details about 2NX can be found in a blog post and the product page.

PowerVR Series9XE and 9XM GPUs

Click to Enlarge

The Series9XE GPU family is an up to the previous generation Series8XE family with the same fill-rate density, but improved application performance of up to 20%, with the GPU expected to be used in cost-sensitive products such as digital TVs, set-top boxes, streaming sticks/dongles, and entry-level to mid-range mobiles and tablets.

The Series9XM family improve performance by up to 50% over the Series8XEP family with increased compute density, and should be found in premium set-top boxes, mid-range smartphones, tablets and automotive ADAS applications.

Both families benefit from improvements in the memory subsystem, reducing bandwidth by as much as 25%, include a new MMU, standard support for 10-bit YUV, and are suitable for 4K output/displays.

Some of the key benefits of the new Series9XE/9XM family include:

  • Performance/mm2
    • 9XE GPUs’ improved gaming performance while maintaining the same fillrate density compared to the previous generation
    • 9XM GPUs’ several new and enhanced architectural elements enable up to 70% better performance density than the competition (as of August 2017), and up to 50% better than the previous 8XEP generation
  • Bandwidth savings of up to 25% over the previous generation GPUs through architectural enhancements including parameter compression and tile grouping
  • Memory system improvements: 36-bit addressing for improved system integration, improved burst sizes for efficient memory accesses, and enhanced compression capabilities
  • Low power consumption thanks to  Imagination’s Tile Based Deferred Rendering (TBDR) technology
  • Support for hardware virtualization and Imagination’s OmniShield multi-domain security, enabling customers to build systems in which applications and operating systems can run independently and reliably on the same platform
  • Support for Khronos graphics APIs: OpenGL ES 3.2, and Vulkan 1.0
  • Support for advanced compute and vision APIs such as RenderScript, OpenVX 1.1 and OpenCL 1.2 EP
  • Optional support for PVRIC3 PowerVR lossless image compression technology

The company also explains the Series9XE/9XM GPU are ideal for use with the new PowerVR 2NX Neural Network Accelerator, which means NNA’s will not only be found in premium devices, but also in entry level and mid range products.

The IP is available for licensing now with four Series9XE GPU IP cores:

  • 1 PPC with 16 F32 FLOPS/clock(GE9000)
  • 2 PPC with 16 F32 FLOPS/clock (GE9100)
  • 4 PPC with 32 F32 FLOPS/clock (GE9210)
  • 8 PPC with 64 F32 FLOPS/clock (GE9420)

and three Series9XM GPU IP cores:

  • 4 PPC with 64 FP32 FLOPS/clock (GM9220)
  • 4 PPC with 128 FP32 FLOPS/clock (GM9240)
  • 8 PPC with 128 FP32 FLOPS/clock (GM9240)

Visit the product page for more details about the new PowerVR GPU families.

Review of Vobot Alarm Clock with Alexa

September 17th, 2017 3 comments

Karl here with a review of Vobot sent By Cafago. I had to Google it when I was asked to review it. Turned out it was an Echo type device with a pixel display and a battery. It started as an Indiegogo campaign. I had been wanting to try to do some sort of voice control with my home automation so I agreed to review it.

Vobot Clock C1 Specifications

These are pulled from Vobot’s website. No power supply is included but a long USB cable is.

Click to Enlarge

Vobot Setup

I let my wife do the initial setup as I figured that she would use it the most. She followed the instructions, and it seemed straightforward from what she told me. She said she had to reboot it once during a step but it continued the setup with no problems. She tied to our Amazon Prime account, and she quickly was playing some music. During research, I did find out that it was not an always listening device.

Firmware Update

I logged into today to see if anything had changed, and there was an update. It suggested that I rebooted the device so I did before updating. Without logging in I wouldn’t have known there was an update. Maybe I missed something but I don’t remember seeing or hearing some sort of notification. I received a verbal notification that it could take up to 10 minutes, but only took a few minutes. The thing is I have no idea what has changed or improved. There is no changelog.

Vobot Display

Display settings allow you to set Brightness, and the time to display Time, Date, Day of Week, Battery Status, and Date + Time.

Click to Enlarge

That’s what the time display looks like.

After pressing the mic button

Get this at times and the eyes blink

Date and time

Hard to catch this one..Starting to play music

Unplugging power and of course get different one when plugging in

Loading music stream


I wanted to do a teardown when I first received the speaker, but I was afraid to break it. Now here at the end, I finally put some force behind it and finally got it apart. Only a few minor scratches and it seems to be fine. The teardown reveals that it is running on a Mediatek MT7688AN, and confirms battery’s capacity. 512MB NANYA storage NT5TU32M16FG-AC completes the list of the main chips. Maybe some enterprising soul will hack this and bring some imaginative new usage.

One big issue

Everything that I tried worked the way I expected for the most part . Home assistant can emulate a Hue bridge, but after reading in the forums, it only works with an Amazon Echo or Google Home speaker. Bummer, that is not the real issue I wanted to bring up. I really wish that it had an always listening microphone. You have to press the button to put it in listening mode. I understand that it has a battery, and would drain the battery but why not have it always listening when plugged in and use the button when roaming about.

Random final thoughts

OK now that I have a device that will take voice commands now what. I like the scrolling display. It is pretty cool and a little retro. I used this about 95% of the time just to play music. Don’t expect much from the speaker, but you can send audio to a home system through the 3.5mm jack on the back.

I know that there are 1000’s of skills but few attracted me. I did like the idea of calling another Alexa device but not supported. Arggh, OK maybe another issue. I did use the weather feature asking about the weather for the next day on occasion.

My 5 year old son was easily able to start music, and it could understand his voice which surprised me. The display is nice, and battery powered is a plus, but I don’t understand one thing. For just about the same price, I can get an Echo Dot which gets me always listening, and 100% works with all the features but no battery or display. I bet that the limitations with the exception of the always listening is inherent to all non Echo devices.

If you are looking for a portable Alexa powered device with a display then the Vobot might be for you. Seems sturdy. Descent battery life. I listened for about 2 hours and it still had a charge on the battery. To get an official Echo Tap it sets you back $120. It is the only official Echo that has a battery.

I would like to thank Cafago for sending the device for review. They provided a coupon code “V3127SA” for the Vobot which is good until 9/30/17, and brings the price down to $ 41.99/€36.1. You’ll also find it for $45 and up on other sites such as or Amazon.

Arm Research Summit 2017 Streamed Live on September 11-13

September 11th, 2017 2 comments

The Arm Research Summit is “an academic summit to discuss future trends and disruptive technologies across all sectors of computing”, with the second edition of the even taking place now in Cambridge, UK until September 13, 2017.

Click to Enlarge

The Agenda includes various subjects such as architecture and memory, IoT, HPC, computer vision, machine learning, security, servers, biotechnology and others. You can find the full detailed schedule for each day on Arm website, and the good news is that the talks are streamed live in YouTube, so you can follow the talks that interest you from the comfort of your home/office.

Note that you can switch between rooms in the stream above by clicking on <-> icon. Audio volume is a little low…

Thanks to Nobe for the tip.

Amlogic A111, A112 & A113 Processors are Designed for Audio Applications, Smart Speakers

September 9th, 2017 6 comments

Amlogic processors are mostly found in TVs and TV boxes, but the company is now apparently entering a new market with A111, A112, and A113 audio processors. I was first made aware of those new processors through Buildroot OpenLinux Release Notes V20170831.pdf document posted on their Open Linux website, where two boards with Amlogic A113D and A113X are shown.

S400 Version 03 Board

First, S400 board with the following key features/specifications:

  • SoC – Amlogic A113D CPU
  • System Memory – 1GB DDR3
  • Storage – 512MB SLC NAND flash
  • Display I/F – MIPI interface
  • Connectivity – Gigabit Ethernet SDIO WiFi/BT (AP6356S)
  • Audio
    • 2x Audio headers (MIC_Connector & SPK_Connector)
  • USB – 1x USB 2.0 OTG
  • Expansion – 2x PCIe ports
  • Misc – 6x ADC Keys, IR_IN/IR_OUT, UART Interface (RS232)

The second S420 board is based on A113X SoC, and comes with less features (no display, no Ethernet, no PCIe…), less memory:

  • SoC – Amlogic A113X CPU
  • System Memory – 512 MB DDR3
  • Storage – 512MB SLC NAND flash
  • Connectivity – SDIO WiFi/BT (AP6356S)
  • Audio
    • SPDIF_IN
    • 2x Audio headers (MIC_Connector & SPK_Connector)
  • USB – 1x USB 2.0 OTG
  • Misc – 6x ADC Keys, IR_IN/IR_OUT, UART Interface (RS232)

The document also explains how to build Linux built with buildroot (you’ll need an Amlogic account), and use audio via applications or frameworks such as aplay, gstreamer, alsaplayer, shairport (Airplay), VLC, DLNA, etc…

Information about Amlogic A113X/A113D processor is lacking on the web, but I eventually found that Amlogic had a YouTube account with now a whopping two subscribers (including yours truly), and one of the two videos was an Alexa voice services demo on Amlogic A113 with what looks like a microphone array inserted on the top of the board.

Further research led me to a page in Chinese discussing Amlogic A111, A112, A113 audio processors, and revealing that Xiaomi AI smart speaker is based on Amlogic A112 quad core Cortex A53 processor, that also shows up in GeekBench running Android 6.0. They also report that A113 features the same four Cortex 53 cores, but has better audio capabilities with 8x PDM interfaces, and 16x I2S interfaces. I also found a page about a microphone array designed for Amlogic S905/S912/A112, and based on Knowles SPH0645LM4H-B miniature microphones .

Finally, I decided to go directly to Amlogic website, and they do have pages for A111 and A112 SoCs, strangely not indexed by search engines so far.

Amlogic A111 key features:

  • CPU – Quad-core ARM Cortex-A5
  • Audio Interface
    • 2-channel I2S input and output
    • TDM/PCM input and output, up to 8 channels
    • S/PDIF output
  • Video Interface – LVDS and MIPI-DSI panel output
  • Security – Supports secure boot and secure OS
  • Ethernet – 10/100/1000M MAC
  • IP License (Optional) – Dolby Digital, Dolby Digital Plus, DTS Digital Surround, DTS HD, DTS Express
  • Process – 28nm HKMG

Amlogic A112 key features:

  • CPU – Quad-core ARM Cortex-A53
  • Audio Interface
    • 8-channel I2S and S/PDIF input and output
    • TDM/PCM input and output, up to 8 channels
    • 2-channel PDM input
  • Video Interface – RGB888 output
  • Security – Supports secure boot and secure OS
  • Ethernet – 10/100M MAC+PHY
  • IP License(Optional) – Dolby Digital, Dolby Digital Plus, DTS Digital Surround, DTS HD, DTS Express
  • Process – 28nm HKMG

If you are interested in evaluating / playing with those processors, and cannot get hold of Amlogic boards (since they only deal with companies), one solution is to get Xiaomi AI smart speaker available for pre-order/arrival notice on sites likes GearBest or GeekBuying, and expected to ship on October 1st.

Thanks to vertycall for the tip.

Huawei Introduces Kirin 970 Mobile SoC with Built-in Neural Processing Unit for Artificial Intelligence Applications

September 3rd, 2017 10 comments

Huawei has unveiled their latest Kirin 970 SoC for premium smartphone at IFA 2017. The processor features an 8-core CPU, and a new 12-core GPU, but what makes this new processor stands apart it a dedicated Neural Processing Unit (NPU) to accelerate tasks used for artificial intelligence, with the company claiming the NPU delivers 25 times the performance with 50x greater efficiency compared to a quad core Cortex A73 processor for AI computing tasks. In practical terms, Kirin 970 could process 2,000 images per minute in an unnamed benchmark image recognition test.

Image Source: Roland Quandt

The press release also mentions the processor will pack 5.5 billion transistors packed into 1 cm² and be manufactured using a 10nm “advanced processor”. But they did not go into much further details about the specs, so instead, I used info from Anandtech and the slide above to derive Kirin 970 technical specifications:

  • CPU – 4x ARM Cortex A73 cores @ up to 2.4 GHz, 4x ARM Cortex A53 cores @ up to 1.8 GHz
  • GPU – 12-core Mali G72MP12 GPU
  • NPU – Kirin NPU delivering up to 1.92TFLOPS (FP16)
  • DSP – Image DSP with 512-bit SIMD
  • Sensor Hub – i7 Sensor Processor
  • Memory I/F  – LPDDR4X @ 1866MHz
  • Storage I/F – UFS 2.1
  • Video – 4K video decoding with HDR @ 60 fps, 4K video encoding @ 30 fps
  • Audio – 32-bit codec, 384 Kbps
  • Camera – Dual camera ISP with face & motion detection
  • Cellular Connectivity – Worldmode LTE Cat 18 with up to 1.2 Gbps download; Dual LTE SIM card support
  • Security – InSE and TEE security engines
  • Manufacturing Process – 10nm

Huawei also uploaded a promo video making clear in no-uncertain terms that the new processor will transform your phone into a digital version of the overly attached girlfriend who “is always here in real-time”, “only for you”, “ready to learn even more” (about you)…

Huawei Mate 10 will be the first smartphone based on Kirin 970 processor, and will launch on October 16th in Munich, Germany.

Google Assistant News – AIY Voice Kit For Sale, Offline Support, 3rd Party Smart Speakers Announced

September 1st, 2017 5 comments

There’s been a lot of development related to Google Assistant in the last few days. First, Google provided an update for AIY Projects, with their AIY Projects Voice Kit now available for pre-order on Micro Center for $35 including a Raspberry Pi 3 board, making the kit virtually free, although you may also purchase it. Note that Micro Center blocks traffic originating from some countries, so I had to use Zend2 to access the site. [Update 10/09/2017: You can also get it from Seeed Studio for worldwide shipping]

Click to Enlarge

Google also announced the Speech Commands Dataset with 65,000 one-second long utterances of 30 short words, which they are in the process of integrating with the next release of the Voice Kit, and will allow the devices to respond to voice commands without the need for an Internet connection. So if you lose your Internet connection, or want to isolate your Voice Kit from it, you can still perform simple tasks like turning on/off lights without an Internet connection.

In this first blog post, the company also showcased some projects based on the Voice Kit, and encouraged the community to provide input for the next version with, or showcase your work on social networks using #AIYprojects hash tag.

The next day, Google published another blog post explaining Google Home, eligible Android phones, iPhones, Google Allo and Android Wear, will soon be joined  by third party speakers supporting Google Assistant and supporting the same features like answering requests, playing music, and controlling appliances. One day later a bunch of announcements was made at IFA 2017, and the company updated their blog post with some list of 3rd party Google Assistant Speakers all scheduled to launch by the end of the year, or early 2018:

  • JBL LINK 10, LINK 20 and LINK 300 respectively 8, 10 and 50 Watts WiFi smart speakers coming to UK, Germany and France starting fall 2017 for 169 Euros, 199 Euros and 299 Euros.

  • Onkyo Smart Speaker G3 (VC-GX30) “acoustic suspension” speaker with 80mm pressed-pulp diaphragm woofer, and 20mm soft-dome tweeter. Available in Black of White.

  • Sony LF-S50G with clock showing through speaker, and to be sold for $199.99.

Availability will be limited to some countries only, likely partially due to a lack of language support, with most expected to be available in the US, UK, Australia, Canada, Germany and France.