> I also find it suspicious that this article emphasizes how much buy-in LTE Direct has from gigantic corporations
The last line of the article is the submarine:
"Wireless carriers might even gain a new stream of revenue by charging companies that want to offer services or apps using the technology, Qualcomm says."
The main motivation of this sort of technology transfer program isn't the license revenue itself. That's insignificant when compared to their operating budget. The point of this program is to provide evidence that the research has commercial applications. With this evidence, they are able to go to Congress and show their research helps the economy, which is used as part of the justification of their research budget.
Tesseract is really difficult to tune such that you get decent results from camera images instead of cleanly scanned text.
Ocropus is really slow, but does seem to work better than tesseract out of the box. It's still quite bad with camera-captured images and IIRC doesn't give you any hooks to improve the results.
I couldn't find any other viable options when I was looking about six months ago.
I want an SDK that gives me an easy way to train the OCR engine. That means being able to view the images which performed badly, with text bounding boxes and the OCRed text overlaid that allow me to manually correct the results which it will then learn from. And, perhaps most importantly, good defaults for the common cases of document scans & camera captures.
The challenge is not technical. The challenge is community. RapGenius is a network effects driven site, so it's utility is driven by how many people use the site. Technology is table-stakes, but branding and marketing matter much more at this point.
RapGenius grew because their user experience was much better than their competition. This was the foundation for their branding & marketing. Their growth will be sustained by their content and (once these penalties are phased out) by users' ability to discover their unique content. The technology is merely an enabler for the community-generated content. New competitors must find ways around the community's cold-start problem.
I largely agree with the article, but also agree with your conclusions.
>Instead, the "competition" is by delivering Rihanna for free by infringing copyrights.
This is a problem for Rihanna and a Maserati problem once a musician become famous, but the author states this article does not apply to "the majors or rock stars."
The challenge faced by the indies is as old as the music industry, they have merely been exacerbated by technology. Even before the decline of music sales, the main challenge for an indie musician was discovery. I think we have a confirmation bias about the past. There have always been unsuccessful musicians, but we forget they existed. Starving artist is not a new cliche.
Have people rejected indie music or is there something else behind superstar popularity? Conventional thinking is that free music makes it easier to get discovered (but harder to monetize, once discovered) because it cost nothing for consumers to try new music. It turns out that there is so much content/competition out there now (because music cost less to produce + free distribution), consumers have choice overload. This makes it harder for people to try indie music.
Music is cultural, so what is popular follows the same power-law dynamics as social networks (in fact, it is driven by social networks). Hence, what's popular becomes more popular.
Agree that the fact that the product has a zero marginal cost does not mean people are unwilling to pay. The entire software industry is built upon a business model of building a product with high production costs and zero marginal cost.
This "dropped phone" search in interesting. This could be the most dangerous aspect of this program.
Chances are little scientific research was done on the accuracy. What is the false positive rate? Are there policies and procedures in place to determine if the phones actually matched the individual? This can't be done without the voice data.
Are the analysts savvy enough about the limitation of the underlying algorithms?
If these questions have not been addressed, it's possible innocent people have been prosecuted.
A big topic of discussion in music tech right now is active vs passive usage. Active usage is when someone uses a music service or station with the intent of discovering new songs and artists. This mode of consumption values a diverse playlist.
On the other hand, passive consumption is when the user just wants background music and doesn't care what's playing as long as it sounds good and matches the mood. This mode tends to favor a more limited playlist, as the overall quality of the music will be higher.
http://www.popsci.com/cars/article/2004-09/obsession-mr-sing...