Which Is the Winner: Google Speech or Amazon Transcribe?
October 22, 2018
The answer appears in a write up called “Google Speech Vs Amazon Transcribe: The War Of Speech Technology.” Spoiler: It looks like Google stomps Amazon in this remarkable analysis.
When it comes to languages supported, Google is king; 119 to Amazon’s paltry dozen or so.
Which system does better with long talking, it is a tie. Two hours is the audio limit. But the write up does not explain if the two hours are chunked or if multiple two hour blocks can be strung together. Important information excluded or ignored in this head to head analysis.
Programming languages? Looks like a toss up. But Amazon supports Dot Net, which may be important at clients like the US Department of Defense.
Privacy? Yeah, well, not much information. The write up ignores Amazon’s federation capabilities and its cross correlation analytics. And Google? Yeah, advertising. We know what that means.
Audio formats? Google seems to cover the bases. The write up ignores the Amazon APIs, which is not surprising because none of Amazon’s law enforcement and policeware capabilities seem to have hit the author’s radar.
Vocabulary? Well, if Google supports more languages, whatever vocabulary tweaking it permits will crush Amazon’s capabilities.
Additional features? Google can do emotions. Right, Alexa cannot it seems. The Beyond Search goose thinks the author should check into Alexa’s capabilities with regard to Amazon Sagemaker. Just a thought: You, gentle reader, may want to check this out as well.
Wrap up?
Both are equal. But the way the write up presents information, Google is the implicit winner.
The Beyond Search goose happens to disagree. What are those Alexa enabled gizmos doing? Interesting question which this Analytics India article does not address.
Stephen E Arnold, October 22, 2018
Amazon: Device Proliferation and One Interesting Use Case
September 21, 2018
The technology “real news” channels are stuffed with Amazon gizmo news. Interesting stuff if one considers that these devices may snap into the eCommerce company’s policeware subsystems.
Here in Harrod’s Creek, we noted one announcement almost lost in the flood of device announcements. “Skype Calling Coming to Amazon Alexa Devices Later This Year” indicates that the tension between the two companies may be lessening. Years ago Microsoft had database envy generated by the eCommerce giant’s innovations in data management and data wrangling. Then there were the skirmishes over staff and office space.
If the information in the ZDNet “real news” write up is accurate, this statement may be more interesting than using an Alexa gizmo as a telephone:
Alexa users will be able to make outgoing Skype voice and video calls, accept incoming Skype calls and make SkypeOut calls to most phone numbers around the world, according to Microsoft officials. Users will be able to say “Alexa, call Jimmy on Skype,” or to say “Alexa, answer” when a Skype call comes in.
But the “real news” continues with an admission from the author:
I have to admit at this point I am kind of lost as to how Microsoft hopes to differentiate and position Cortana. Granted, Microsoft execs said they want Cortana not to be just about convenience, but about built-in assistance, but Skype is a Microsoft service….
From my vantage point in Harrod’s Creek, the tie up in voice may be more than a test. In fact, the deal may signal another victim of the Amazon strategy. Microsoft may be losing without knowing that it is in a fight.
Stephen E Arnold, September 21, 2018
Amazon Joins Visual Search Parade
July 30, 2018
Text search is long past done as a frontier. Verbal search is already being nailed down by more startups and tech giants than you can tell Alexa to shake a stick at. So, the new frontier? It’s visual search and, you guessed it, one of the biggest names in the industry is already working their way in, as we discovered in a recent Fortune story, “Snapchat and Amazon are Working On Visual Search Feature.”
According to the story:
“Snap appears to be laying the groundwork for a partnership with e-commerce giant Amazon. “According to TechCrunch, a version of Snapchat being developed for Android phones includes code for a new feature called “Visual Search” that can use Snapchat’s camera to send images of a product or a barcode scan to Amazon, which then display search results.”
Amazon is not alone, however. Microsoft is also developing a visual search tool that can simply look at items and begin shopping for them. The controversy about the accuracy of Amazon’s Rekognition system may inhibit some of Amazon’s plans for image centric features and functions. I I search for a product with my mobile phone and Amazon returns matches which are incorrect, what happens to consumer confidence?
Error rates are likely to matter, probably more when looking for a shirt than when trying to figure out which elected official is a bad actor. Shirts are different. Bad actors not so much, some may suggest.
Patrick Roland, July 30, 2018
Facebook: A New York City-Sized PR Problem
July 20, 2018
I read “Once nimble Facebook Trips Over Calls to Control Content.” If you are looking for this write up online, the story’s headline was changed to “What Stays on Facebook and What Goes? The Social Network Cannot Answer.” You may be able to locate the online version at this link. (No promises.) The dead tree version is on Page A1 of the July 20, 2018, edition which comes out on Thursday night. Got the timeline square?
I wanted to highlight a handful of comments in the “real” news story. Here we go with direct statements from the NYT article in red:
- The print version headline uses the phrase “once nimble.” Here in Harrod’s Creek that means stumbling bobolyne. In Manhattan, the phrase may mean something like “advertise more in the New York Times.” I am, of course, speculating.
- I marked in weird greenish yellow this statement: “Facebook still seems paralyzed over how to respond.” So much for nimble.
- Another: “Comically tripped up”. Yep, a clown’s smile on the front page of the NYT.
- My favorite: The context for being a bit out of his depth. Whatever does “yet lucidity remai9ned elusive.” Does this mean stupid, duplicitous, or something else?
- I thought Silicon Valley wunderkind were sharp as tacks. In the NYT, I read “Facebook executives’ tortured musings.” Not Saturday Night Live deep thoughts, just musings and tortured ones at that.
- How does Facebook perceive “real” journalism? Well, not the way the NYT does. I circled this phrase about Alex Jones, a luminary with some avid believers one mine drainage ditch down the road a piece which is Kentucky talk for “some”: “Just being false doesn’t violate community standards” and “Infowars was a publisher with a ‘different point of view.’”
- This is a nifty sequence crafted to recycle another “real” journalist’s scoop interview with Mark Zuckerberg: “what Facebook would or would not allow on its site became even more confusing.” So, a possible paralyzed clown who lacks lucidity is confusing.
- The “bizarre idea” word pair makes sure I understand what the NYT believes in a lack of clear thinking.
But these brief rhetorical flourishes set up this statement:
A Facebook spokeswoman [who is not identified] explained that it would be possible, theoretically, to deny the Holocaust without triggering Facebook’s hate-speech clause.
Those pesky algorithms are at work. But the failure to identify the person at Facebook who offered this information is not identified. Why not?
Here’s another longer statement from the NYT write up:
And what exactly constitutes imminent violence is a shifting line, the company said— it is still ‘iterating on’ its policy, and the rules may change.
I don’t want to be too dumb, but I would like to know who at the company offered the statement. A company, to my knowledge, cannot talk unless one considers firing a question at Amazon’s Alexa.
I put an exclamation point on this statement in the NYT article:
All of this fails a basic test: It’s not even coherent. It is a hodge podge of declarations and exceptions and exceptions to the exceptions.
Net net: Facebook has a public relations problem with the New York Times. Because of the influence of the “real” newspaper and its “real” journalists, Facebook has a PR problem of magnitude. Perhaps the point of the story is to create an opportunity for a NYT ad sales professional to explain the benefits of a full page ad across the print and online versions of the New York Times?
Stephen E Arnold, July 20, 2018
Journalists: Smart Software Is Learning How to Be a Real Journalist
July 15, 2018
I read “Why Bots Taking Over (Some) Journalism Could Be a Good Thing.” I love optimists who lack a good understanding of how numerical recipes work. The notion of “artificial intelligence” is just cool like something out of science fiction like “Ralph 124C 41+” except for the wrong headed predictions. In my 50 year work career, technologies are not revolutions. Technologies appear, die, reform, and then interact, often in surprising ways. Then one day, a clever person identifies a “paradigm shift” or “a big thing.”
The problem with smart software which seems obvious to me boils down to:
- The selection of numerical recipes to use
- The threshold settings or the Bayesian best guesses that inform the system
- The order in which the processes are implemented within the system.
There are other issues, but these provide a reasonable checklist. What does on under the kimono is quite important.
The write up states:
If robots can take over the grunt work, which in many cases they can, then that has the potential to lower media organizations’ costs and enable them to spend a greater proportion of their advertising income on more serious material. That’s terrible news for anybody whose current job is to trawl Twitter for slightly smutty tweets by reality TV show contestants, but great news for organizations funding the likes of Guardian journalist Carole Cadwalladr, who broke the Facebook / Cambridge Analytica scandal. Isn’t it?
Good question. I also learned:
Technology can help with a lot of basic reporting. For example, the UK Press Association’s Radar project (Reporters And Data And Robots) aims to automate a lot of local news reporting by pulling information from government agencies, local authorities and the police. It’ll still be overseen by “skilled human journalists”, at least for the foreseeable future, but the actual writing will be automated: it uses a technology called Natural Language Generation, or NLG for short. Think Siri, Alexa or the recent Google Duplex demos that mimic human speech, but dedicated to writing rather than speaking.
I recall reading this idea to steal:
In fact, human reporters will continue to play a vital role in the process, and Rogers doesn’t see this changing anytime soon. It’s humans that make the decision on which datasets to analyze. Humans also “define” the story templates – for example, by deciding that if a certain variable in one region is above a particular threshold, then that’s a strong indicator that the data will make a good news story.
Now back to the points in the checklist. In the mad rush to reduce costs, provide more and better news, and create opportunities to cover certain stories more effectively, who is questioning the prioritization of content from an available stream, the selection of items from the stream, and the evaluation of the data pulled from the stream for automatic story generation?
My thought is that it will be the developers who are deciding what to do in one of those whiteboard meetings lubricated with latte and fizzy water.
The business models which once sustained “real” journalism focused on media battles, yellow journalism, interesting advertising deals, and the localized monopolies. (I once worked for such an outfit.)
With technology concentration a natural consequence of online information services, I would not get too excited about the NLG and NLP (natural language generation and natural language processing services). These capabilities for smart software will arrive. But I think the functionality will arrive in dribs and drabs. One day an MBA or electrical engineer turned business school professor will explain what happened.
What’s lost? Typing, hanging out in the newspaper lunch room, gossip, and hitting the bar a block from the office. Judgment? Did I leave out judgment. Probably not important. What’s important that I almost forgot? Getting rid of staff, health coverage, pensions, vacations, and sick leave. Software doesn’t get sick even though it may arrive in a questionable condition.
Stephen E Arnold, July 15, 2018
Amazon Factoid: Home Speaker Department
July 9, 2018
Short honk: I read “What Cracking Open a Sonos One Tells Us aboiut the Sonos IPO.” In the write up was an interesting to me item of information. Here is what I noted:
Even though both [Sonos and Echo Plus] of these products are different in pretty much every decision that was made surrounding the hardware, they use the same backend Alexa service (where most of the IP is) from Amazon.
Interesting. Amazon’s approach allows it to generate revenue from a customer (maybe partner?) and from its own product line.
This appears to be a double dipping approach of value. What happens if Amazon decides to raise its prices for a customer (partner)? I suppose the Sonos-type outfit can hightail it to IBM’s, Google’s, or Microsoft’s cloud.
That may pose costs, timing challenges, and technical hoops. The time required by a Sonos-type outfit might be enough to allow Amazon to shave a few seconds off its lap time.
With Google slashing prices for its home gizmos, the home data ecosystem may become more interesting in the months ahead.
Stephen E Arnold, July 9, 2018
Digital Assistants Working Hard to Be More Human
July 8, 2018
Whether you use Alexa, or Siri, or Cortana, or a host of other AI-infused digital assistants, the producers of that technology have something in common: they want those electronic personalities to be more human. Interesting moves are being made in this world to make that happen, according to a recent Inquirer story, “Microsoft Snaps Up Semantic to Make Cortana Seem a Bit Less Robotic.”
According to one Microsoft exec:
“Combining Semantic Machines’ technology with Microsoft’s own AI advances, we aim to deliver powerful, natural and more productive user experiences that will take conversational computing to a new level.”
The story continued:
“Google, of course, has Duplex which can make natural sounding voice calls on your behalf. It has also suggested it is looking into the idea of giving Assistant a back story.”
However, this comes with a consequence. As Wired pointed out, as these assistants get more comfortable with inflection and reading our voices, the opportunity for manipulation becomes eerily more present. These near-human tools are not to that point yet, but we don’t doubt that it’ll arrive soon. Who wants to type when one can talk, think a human is on the other end of the connection, and be so much more efficient.
Patrick Roland, July 10, 2018
Calendars Are Now Search… If One Is Busy and Eschews Print Schedulers
July 3, 2018
You might not think it, but your doctor’s appointments and dinner parties are a big deal to search companies. With the rise of digital assistants like Siri and Alexa, your datebook is the next big horizon to conquer. The ways in which this will unfold might surprise you, according to a recent Japan Today story, “Google’s ‘Reserve’ Tool Winning Converts and Taking Search to the Next Level.”
According to the story:
“[S]even software firms that supply schedule data to Google described the volume as significant, with as much as 75 percent of bookings representing new customers. Consumers like the convenience. Business owners say the tool is putting their names in front of more potential clients.”
It is no coincidence that several experts are touting the ability of digital assistants to help with travel planning. In a weird way, voice search can now do a lot of the work of a travel agent, in terms of eyeing your schedule, finding deals, and even purchasing flight tickets. From getting reservations to booking flights to making sure someone is picking up junior from soccer practice, there is a revolution happening in search and how it relates to daily life. Search and scheduling: A wonderful way to fill one’s day with useful activities.
Patrick Roland, July 3, 2018
IBM Demo: Debating Watson
June 29, 2018
IBM once again displays its AI chops—SFGate reports, “IBM Computer Proves Formidable Against 2 Human Debaters.” The project, dubbed Project Debater, shows off the tech’s improvements in mimicking human-like speech and reasoning. At a recent demonstration, neither the AI nor the two humans knew the topics beforehand: space exploration and telemedicine. According to one of the human participants, the AI held its own pretty well, even if it did rely too much on blanket statements. Writer Matt O’brien says this about IBM’s approach:
“Rather than just scanning a giant trove of data in search of factoids, IBM’s latest project taps into several more complex branches of AI. Search engine algorithms used by Google and Microsoft’s Bing use similar technology to digest and summarize written content and compose new paragraphs. Voice assistants such as Amazon’s Alexa rely on listening comprehension to answer questions posed by people. Google recently demonstrated an eerily human-like voice assistant that can call hair salons or restaurants to make appointments…But IBM says it’s breaking new ground by creating a system that tackles deeper human practices of rhetoric and analysis, and how they’re used to discuss big questions whose answers aren’t always clear. ‘If you think of the rules of debate, they’re far more open-ended than the rules of a board game,’ said Ranit Aharonov, who manages the debater project.
The demo did not declare any “winner” in the debate, but researchers were able to draw some (perhaps obvious) conclusions: While the software was better at recalling specific facts and statistics to bolster its arguments, humans brought more linguistic flair and the power of personal experience to the field. As for potential applications of this technology, IBM’s VP of research suggests it could be used by human workers to better inform their decisions. Lawyers, specifically, were mentioned.
Keep in mind. Demo.
Cynthia Murrell, June 29, 2018
SoundHound Is Ready to Compete
June 22, 2018
What began as a music-identification tool is now a platform for building one’s own voice assistant application. Business Insider reports, “This 13-Year-Old Startup Just Got $100 Million and Is Valued at Over $1 Billion—Now It’s Taking on Amazon, Google, and Apple.” Writer Kif Leswing notes this recent funding round is led by Chinese firm Tencent, with other notable contributors like Hyundai, Daimler, and Europe’s Orange. We learn:
“SoundHound did not disclose its valuation following this round, but a person familiar with the company says it’s worth over $1 billion, making it a unicorn. It’s going to use those funds to go up against some of the biggest names in technology, including Apple, Amazon, and Google. SoundHound CEO Keyvan Mohajer tells us that he’s not afraid of those 800-pound gorillas, though. ‘We said, don’t be afraid,’ said Mohajer. ‘I always tell my team members, think of your competitors are variables in a complex set of equations.’”
That’s one way to look at it. Mohajer figures his company has one important advantage over those huge players—a less obtrusive integration into clients’ businesses. By creating, and branding, their own voice assistants, companies retain more control over their image and keep customers focused on them throughout the user experience.
Leswing notes that much of SoundHound’s funding has come from strategic partners, as opposed to financial firms. This is because, he reports, Mohajer is gathering allies in his face-off against the likes of Amazon. Mohajer declares he had to turn investors away for this funding round, so there does seem to be abundant interest in Houndify. Should Alexa be worried?
Cynthia Murrell, June 22, 2018
-
- Subscribe to Beyond Search
Feature archive
News archive
-
