It was a big deal last year when Google came out with its own smartphone, the Pixel. Also one year ago, the tech giant released a smart home speaker and debuted the AI-powered Assistant that lives inside these new gadgets. Now it's twelve months later, and Google is stepping on the gas. This week, we saw a new Android flagship phone, the Pixel 2. Also, two new Google Home speakers, a new Chromebook with a button to call the Assistant, some wireless earbuds called Pixel Buds with tap-to-talk functionality for Assistant, an AI-powered camera called Clips. Google is now truly a hardware player, but it has a powerful software engine running across all of its products.

Assistant was launched a year ago during Google's first hardware event. During the company's Pixel 2 unveiling yesterday, Google said Assistant is meant to be "helpful, simple to use [and] available where you need it." With these fundamentals in mind, Assistant will soon be expanded with several new features meant to differentiate it from competitors including Siri and Alexa. Running Routines Amazon recently unveiled its new range of Alexa-enabled products. They come with Routines, a feature that allows several linked actions to be run from a single voice command.

Harman Kardon's Cortana-powered speaker might be hitting stores sometime soon. The Verge reports that a listing for the Invoke with Cortana has shown up in Microsoft's store alongside a $200 price tag. We got a peek at the speaker in May at Computex 2017 though we haven't been able to experience its capabilities for ourselves quite yet. Along with the Invoke, Harman Kardon recently announced three new Google Assistant-powered JBL Link speakers as well as an Alexa-enabled speaker dubbed the Allure -- completing its virtual assistant trifecta.

Apple, Google and Amazon have all announced smart speakers that are more like gadgets than audio gear. Of course, whenever a tech company attempts to build audio gear, the specter of the iPod HiFi looms large. Apple's ill-fated attempt to build the perfect speaker was considered a flop thanks to its high price and limited functionality. It was a glorified iPod dock that set you back $349 and couldn't even connect to the internet or FM radio. A decade later, smart speakers look ready to take over the world.

To understand why WaveNet improves on the current state of the art, it is useful to understand how text-to-speech (TTS) - or speech synthesis - systems work today. The majority of these are based on so-called concatenative TTS, which uses a large database of high-quality recordings, collected from a single voice actor over many hours. These recordings are split into tiny chunks that can then be combined - or concatenated - to form complete utterances as needed. However, these systems can result in unnatural sounding voices and are also difficult to modify because a whole new database needs to be recorded each time a set of changes, such as new emotions or intonations, are needed. To overcome some of these problems, an alternative model known as parametric TTS is sometimes used.

Earlier this week, a small startup called Init.ai announced that it soon would be discontinuing its service -- a smart assistant for customer representatives to parse and get better insights from their interactions with users, as well as automate some of the interactions -- because the team was (according to a notice on the site) "joining a project that touches the lives of countless people across the world." TechCrunch has now learned what that project is: the team is joining Apple. They are becoming a part of the group working on Siri, Apple's own personal assistant that uses voice-based commands and natural language to answer questions, control your phone and more. To be clear, this is not a straight acquisition, but more of an acqui-hire, so Apple didn't have a comment to offer along the lines of the one it made last week. Specifically, I understand there is no IP coming over and (typical of Apple) no specific indication of what part of Init's code or services will be used by Apple, if any.

Google will release a smaller version of its Home smart speaker, the company revealed during an event on Wednesday. The device, called the Google Home Mini, will cost $49 and launches on October 19. Like the larger Google Home, the Mini is partially made of fabric and will be available in multiple colors. There are four LED lights underneath the the fabric that illuminate when the device is listening. Google says the Home Mini will be capable of connecting to any Chromecast compatible speaker and can project audio in 360 degrees.

Google on Wednesday unveiled a new smart speaker called the Home Max, which is designed to intelligently adjust the audio depending on a user's surroundings. If that sounds familiar, it should: Apple's Siri-powered HomePod speaker, which it announced in June, is also intended to adapt to its audio based on its environment. Google says the Home Max will be available for $399 in December, $50 more than Apple's HomePod, which is expected to debut in the same timeframe. The Home Max's sound is 20 times more powerful than that of the standard Google Home, and the speaker uses artificial intelligence to alter the sound accordingly. If one were to move the Home Max a few feet in a particular direction, for example, the speaker would be able to compensate for that change.

If Google made one thing clear at its event on Wednesday, it's that it wants its artificially intelligent virtual assistant to be everywhere, from phones to smart earbuds and laptops. Google unveiled several new products on Wednesday, many of which rely on its Google Assistant to provide compelling features that differentiate them from competing gadgets. Here's a look at everything Google just announced. Google's new Pixel smartphones were the stars of Wednesday's show. The new phones, one of which includes a 5-inch display and the other which has a 6-inch screen, both include improved displays and cameras.

"Data wrangling" was an interesting phrase to hear in the machine learning (ML) presentations at Microsoft Ignite. Interesting because data wrangling is from business intelligence (BI), not from artificial intelligence (AI). Microsoft understands ML incorporates concepts from both disciplines. Further discussions point to another key point: Microsoft understands that business-to-business (B2B) is just as fertile for ML as business-to-consumer (B2C). ML applications with the most press are voice, augmented reality and autonomous vehicles.