A series of recent postings from the BBC have questioned the validity of the SpinVox as a “speech technology company.” The posts (one news item and a blog post) raise questions around whether there’s an actual speech engine powering the SpinVox service or just an elaborate archipelago of call centers flung around the world, transcribing messages by hand, as some have accused.

The BBC cites photos from a Facebook group created by staff at an Egyptian call center, RAYA, which used to work for Spinvox, “containing what appears to be sensitive commercial information” as an indication that the human transcription may play a bigger role than SpinVox would like to let on. The blog especially seems to hint that almost all of SpinVox’s transcriptions are done by hand. The company, for its part, has fired back, calling the BBC’s reporting “incorrect” and “inaccurate,” in its own blog response.

Christina Domecq, SpinVox CEO, goes one step further in her response to the Guardian. Domecq says, “The majority of calls are fully automated.”

I was actually able to find the Facebook group that the BBC mentions, “Sp!nVox R@Y@,” pretty easily. It hasn’t been updated in about a year, which may mitigate its importance since SpinVox claims to be upping the proportion of automation behind its service. That said, among the claims made in the group’s description are: “we work behind the scenes…we are the invisible heroes[sic]… we are Tenzing… we are unknown and they deny our existence[sic] !!”

The photo which BBC alleges might contain sensitive commercial information is pictured below.

“The photo is from a training session – that is a dummy message,” claims Rachael Lyons, director of North American communications for SpinVox, in an email to Speech Tech–in other words, a fake message that isn’t sensitive at all.

“[The RAYA agents] were using training data – a model system that SpinVox uses to evaluate the quality of call centre support before contracting with the supplier to handle real user data,” writes SpinVox on its blog. “The training system will require individuals to convert full messages in order to establish their speed and accuracy.”

“This would not be the case on a live customer system where should the VMCS system need assistance in learning, operators would only be presented with portions of any message for assisted learning,” the blog adds.

To prove the company’s point, Ms. Lyons cites a number of items from the group’s photos:

• Pictures show training screens you can tell that because they have other apps running – doesn’t happen in live systems
• Screens have whole messages on them – doesn’t happen in Live systems
• Pictures show training presentations and materials
• Audio is known training audio we use all the time
• Dates on the site were when SpinVox was testing RAYA
• SpinVox hasn’t required for the site to be taken down because it doesn’t breach security

Subsequent to the trial, SpinVox says that the RAYA call center was not retained to handle live data. It says the same for Kencall in Kenya, which failed to meet criteria and, incidentally, also has a Facebook group that complains about working for SpinVox that also hasn’t seen an update for almost a year.

The BBC’s only named source in the article is Kareem Lucilius who said he worked for six months at the call center, alongside as many as 150 others. He said that after initial training, he went on to transcribe live messages.

“It was done 100% by people,” he claims, adding, “We heard the message from the very beginning to the very end. Love messages, secret messages, messages with sexual content, even people threatening to kill each other.”

Mr. Lucilius, however, seems to have worked for SpinVox under the auspices of RAYA. A Facebook account by that name has commented on the group page (this may be how the BBC reached him) and a “Kareem” is pictured sleeping or pretending to sleep in a photo. If he did work at RAYA, this raises some questions because SpinVox claims that company never saw any live messages.

On the other hand, a post on the Facebook page from April 2008 reads, “Mabrook for those who made it to the Live Session..You guys kick ass!” A second post in June 2008, from the same person, clearly identifying himself as being from RAYA, asks if the UK workers if their servers are down, implying that RAYA may have been processing live messages.

Whatever the situation, it is not helped by the fact that SpinVox’s new and rapidly expanding technology has some secrecy surrounding its center, leaving room for doubts, speculation, and innuendo.

The company won’t say, for instance, how much human intervention is involved in transcription, describing the actual proportion of messages automatically converted as “highly confidential and sensitive.” Rather, it says that it requires only a few hundred agents per market to convert messages without learning assistance. In Argentina, where SpinVox has 10 million customers, it has less than 70 call center staff, suggesting it’s not feasible for SpinVox to operate without some level of automated speech recognition, the Guardian suggests.

Ms. Lyons adds that SpinVox has “reduced human assisted learning to just 2% of what it was when we started.”

What that means without a clearer view of how many messages SpinVox processes with human assitance, however, leaves much to be answered. As my brother Adam B. would say, “This is a real brain buster!”

SpinVox recently ran a survey among Americans, British, and Canadian users of Ping (a service that allows users to update social media via voice through SpinVox’s ASR) to, at least ostensibly, “reveal new slang emerging in the Twittersphere.”

The results produced by the study were the sort just ripe for blogging about. As much as 84 percent of SpinVox’d Twitter users admitted to using the service in public mid-travel, while 23 percent admitted to tweeting from the john, A.K.A. “the W.C.”

Whether or not this means public restrooms and there’s any overlap between that 84 and 23 percent is unclear. SpinVox did not provide a Venn Diagram.

Thirteen percent of users said they used the service on a treadmill, 25 percent jogging in the park, 37 percent while cooking, 17 percent outdoors events like music festivals, 8 percent, and 5 percent waiting in lines.

The “top 10 Twitterverse-releated slang words” (I’ll in the new tradition of making annoying words out of “Twitter” call these “Twitterese”) SpinVox found in the SpinVox “Voxgeist” (C’mon! Really?) are:

• Hashtag (noun): a way of marking an event or common theme e.g. “I’m off to Mobile Geeks of New York #MGoNY”
• #Follow Friday (term – pron. Hash Follow Friday): a weekly event, where you recommend your favorite Twitterers to your followers.
• Dabr (noun): a mobile web interface for Twitter.
• Twetiquette (term): appropriate behavior when using Twitter.
• Twitpoll (noun): a survey/question posed on Twitter, often for research. Sometimes just for fun: e.g.: “Which shoes should I buy?”
• Tweetdeck (noun): one of the most popular desktop interfaces for Twitter.
• Twhirl (noun):  a smaller, less intrusive desktop interface for Twitter.
• Mashable Effect (term): the result of @mashable, aka Pete Cashmore, founder of Mashable.com, re-tweeting your site/blog address and causing it to overload with web traffic.
• Tweetup (noun): a meeting organized on Twitter.
• Twestival (noun): a charity event organized using Twitter

Strangely missing from the list, my brother Adam B. points out, are “speechtacular,” “Speechlandia,” “Speech Head,” and “None of your speechness.”

I have to say, I’m a bit skeptical of how scientifically rigorous this effort was myself. SpinVox hasn’t released a methodology, so no word on sample size or how these figures are derived.

When I asked the company about what kind of standards it was using, Rachel Lyons, SpinVox’s Director of North American Communications, said, “[We] would say the survey was more straw poll than census.”

Hrm.

I’m still willing to give SpinVox the benefit of the doubt here, though. I’m sure their numbers check out, but with questions asking users if they tweet on the can this survey begins to strike me as the sort of thing that is concocted to produce the whimsy and mirth that gets reposted on blogs. The kind of thing that collaterally gets the word out about the fact that you can use SpinVox on Twitter—-a clever P.R. move rather than a question of genuine scientific inquiry. In fact, come to think of it, this study or straw poll or whatever seems suspiciously like a lot of other top ten list based posts that SpinVox has made to be blogged about…

On that count, Speech Heads, I suppose we’re guilty, but why not make yourself complicit in shameless marketing, too? Make sure to follow Speech Tech on Twitter! http://twitter.com/SpeechTech/

Well Speech Heads, this was a long time coming, but here it is! OUR SPINVOX REVIEW!

Some ground rules before we begin:

Pussyfootin’ Provisos and Liability Claims

Trial Version

The SpinVox service I was using was a trial version. It basically worked through call forwarding, sending all my missed calls to SpinVox for transcription and using a third-party aggregator to send them back to me as texts/emails. If I wanted to listen to the messages, I had to call SpinVox directly. When you use SpinVox natively, the service works through a carrier’s existing channels. So for the most part you don’t feel its presence as much as I did.

SpinVox v. Nuance

Doubtlessly, this review will draw some comparisons to our Nuance VM2T review. The devious of you out there will be trying to piece together which one I think is better. Sorry to disappoint, I won’t be coming down with any definitive pronouncements on that count. I’m afraid those betting stubs you bought are going to be worthless.

For the purposes of our blog product reviews we’ve very purposefully eschewed using any kind of numerical value system-in part because that system is just untenable long term for the blog. Technology will change and, moreover, numbers can be used to suggest comparisons between features that we never intended to compare. For a lot of the same reasons, when we do the Speech Industry Awards (SpeechTek 2009, HOLLA!) we have our judges rate vendors overall rather than dealing with thickly overgrown forest individual products.

Furthermore, for the purposes of this Thrilla in Manila: VM2T v. SpinVox, the two are only comparable so far. Like with SpinVox, when I test drove the Nuance service it was just a demo version that didn’t actually deliver texts to my phone. I don’t really have a great feel for the ins and outs of delivery times or the niceties of interface of either. Given that, the only point of comparison between the two that one could feasibly draw is between the respective engine’s recognition accuracy.

The Equipment

Another disclosure: the mobile phone used throughout this review was a Samsung SCH-U540. My phone, pictured here can be described as pretty much the featurest of feature phones. This phone is so feature, and this is not a joke, that when I was talking to a vendor about usability tests on mobile devices, they told me they test on a “full gamut of phones.” At the high end they tested their software some suped BlackBerry, probably capable of fielding a line drive while processing six terabytes of cancer research, and on the low end, they used a phone that the guy called, “as nothing as you can probably get.” The low phone in question? My exact phone.

So bear that in mind when you read through.

And now…

The Review:

Delivery and Some First Results

One of the immediately cool things I noticed about the service is that when I got a message it appeared as a text from the number that called me. Other than “spoken through SpinVox” appearing at the end of each message, they looked exactly like a text from the caller. That makes it really easy to call or text somebody back without having to dig up their number or dial anything. And hey, if the number is in your address book, you’ll know who it’s from before you even crack the message open. Beats cycling through a bunch of messages you have no need to hear, no? Special added bonus, you can also set up your SpinVox account to send a copy to your email!

There were, however, some issues that I encountered in my use of the service. One was with the delay in delivery of messages. How fast they made it from utterance to text seemed to vary anywhere from a couple minutes to as many as fifteen. Some were also dropped while others were delivered in random chronological order. I don’t know how much I can attribute this to my own carrier’s network or SpinVox, but I can provide an example of where things went wrong. (more…)

Speech Heads, it’s a beautiful thing when speech providers compete.

Post our review on Nuance’s VM2T, SpinVox (perhaps a little jealous) wrote me to ask me to the dance—-the speech dance that is. In a matter of days, I’ll begin trying out their service. We’re going to be giving it the full treatment, subjecting it to rigors of my brother Adam B.’s near insane rambling messages, putting it in the wringer of Shakespearean English, pushing it through the Danger Room of the busy New York City streets, and springing a couple of unexpected tests on it.

Full results to come, Speech Heads! Stay tuned! Same speech-time, same speech-channel!

First of all, I know what you’re probably thinking out there in Speechlandia. Where’s this much talked, much huffed about Dragon 10 review that the Brothers B. have been promising? Well, we’ve had to keep mum about this because of an embargo, but we’re finally unfettered. Shortly after we began our review process, we got a phone call from Nuance HQ.

Hold the presses!

They told us that they were going to be releasing 10.1 and did we want to review it? Did we want to review it? Did we want to review it? Sheeyeah we wanted to review it. We were promised some copies and are now just patiently awaiting to begin the process anew with the latest version.

Apparently, the biggest update is that 10.1 is compatible with the 64-bit version of Windows Vista. Likewise, it has a couple of fixes.

But, wait! That’s not all!

(more…)

Just last week, British Speech-To-Text provider SpinVox issued a press release listing its “Top Ten Worst Moments In Public Speaking.”

The list–which includes everyone from George W. Bush to Kate Winslet–features links to news stories about and YouTube clips of the Terrible Moments in Public Speaking.

Also in the release are public speaking tips and a list of the Top 5 Worst Public Speakers: 1. Gordon Brown; 2. David Beckham; 3. Kate Winslet; 4. Chris Moyles; 5. Prince Charles.

Well, Speech-Heads, you like my Speech Brother Eric B., maybe be wondering what this release has to do with Speech Technology, STT, or SpinVox.  All I can say is this: I have no idea.

So,without further ado, here is SpinVox’s Top 10 Worst Moments In Public Speaking:

1. George W.  Bush; “Fool me once”; 2002.

2. Delia Smith; “Lets be ‘avin’ you!”; 2005.

3. Kate Winslet; “Oh, God, who was the other one again?”; 2009 at Golden Globes.

4. Judy Finnigan; “An unfortunate wardrobe malfunction“; 2000 at National TV Awards.

5. Gwyneth Paltrow, “Sobs”; 2005 at Academy Awards.

6. Halle Berry; “Tears and screams”‘; 2002 at Academy Awards.

7. Boris Johnson; “Olympic handover speech“; 2008.

8. Gerald Ratner, “Total crap speech”; 2001.

9. Keven Keegan; “I will love it if we beat them”; 1996.

10. Donald Rumsfeld; “Known unknowns”; 2002 .

Today we cast aside the black veils, cease mourning for Lauren, and resume blogging.

Yesterday, we posted a story about SpinVox’s involvement with MTV’s Staying Alive foundation, which aims to educate young people about sex (we called the story Sex with SpinVox, which we’re going to have to change because when we send it out on our e-newsletter, the word “sex” automatically triggers the spam filters. Context, people!).

While the cause is good, I wonder about the delivery method. A SpinVox spokeman pointed out that “there’s no denying the power of voice on a variety of levels. The idea is to have a spontaneous discussion about sex, sex education, and the issues around HIV, AIDS and other sexually transmitted diseases.” (more…)

For those interested in updating your speech rec parameters, SpinVox had compiled a list of popular US and Canadian slang. The company’s Voxgeist dictionary is tuned into a country’s ever-changing vernacular.

For your enjoyment, here are the Canadian words, ranked in order of popularity, after the jump.

(more…)

Well, I would, if it hadn’t already passed, and I was Catholic. You may remember my announcement last week that I’m testing a new product and will post my reviews on the ST blog. So, which vendor am I reviewing? Hints: it’s related to mobile, free [I <3 beta], and integrates Web 2.0 into its business plan. The full scoop after the jump!

(more…)

It’s time for another edition of the Speech Tech blog’s regular feature, “Crushes & Hexes.” Appearing every Wednesday, we highlight companies, places, technologies, or people we deem praise-worthy, or cringe-inducing within the speech/tech/IT world. As always, your feedback is appreciated! Email us at blog@speechtechblog.com if you have a crush or hex item you’d like to see online.

Crush: Speech Technology in the Mobile Market
Why We’re Loving It: After a few days of uncertainty regarding Nuance and Android, we got a definitive answer yesterday from Mike Thompson at Nuance (see yesterday’s post). The verdict: “We’re actively participating,” Thompson says. Cha-ching! In other Nuance news, the company announced it had signed a multiyear deal with Samsung to incorporate its speech recognition technologies into Samsung mobile phones. But one of the biggest news items was SpinVox’s $200 million-announcement: the company received that much in fundraising rounds with investors like Goldman Sachs. After an impressive showing at the Mobile World Conference in Barcelona, SpinVox has really been hitting it big-time, and with the $200 million, they’ll be doubling the number of mobile providers carrying its voicemail-to-text technology, as well as investing in an Advanced Speech Group in Cambridge.

Hex: Vishing – Giving VoIP and Contact Centers a Bad Name
Why We’re Hating: Following VoiceCon, I kept in touch with Krishna Kurapati, president of voice and information security company Sipera Systems. He sent me a brief email yesterday about “vishing,” or the use of VoIP/Internet phones to trick callers into thinking they’re getting in touch with a call center, when in fact their personal information is being poached by fraudsters. I’m too young to remember, but my mom tells me that people were hesitant to use ATMs when they first hit the scene — I worry that, because VoIP is still relatively new to the call center, that this will make people not afraid of speaking with their company, but also help them learn more about identity protection. In a February 2008 article in the Fort Worth Star-Telegram, writer Teresa Mcusic put it this way:

These so-called “vishing” attempts are on the increase because costs are going down for international phone calls, thanks to Internet phones, and identity-theft call centers are popping up all around the world targeting the U.S.

[Photos courtesy Wikipedia & SecureComputing.com]