Voicing and Mapping
October 28th 2008 08:57
I had a bit of a clean up of baggage and luggage and stuff that was cluttering my office floor yesterday (it being Labour Day) and came across a Netguide magazine that I’d bought earlier in the year but had never really checked out.
A letter to the editor mentioned a couple of sites that sounded interesting, so I’ve just been checking them out. The first is Voxforge.org (not to be confused with Voxforge.com which appears to have been started in 2004 and abandoned). Voxforge’s aim is be a place where you can become part of a service where you offer your voice for free and open source recognition engines.
Voxforge then makes all submitted audio files available under the GPL license, and compiles them into acoustic models for use with Open Source speech recognition engines such as Sphinx, ISIP, Julius and HTK.
GPL, for those, like me, who have never heard of it, is a General Public License. With normal copyright an author has the right to control copies and changes to a work. With a GPL license (also known as "copyleft") a user receives the right to copy and change a work. It’s all part of the process of making the Internet ‘free’, an important word in cyberspace.
Using Voxforge appears to be pretty straightforward: you read and record some text, and upload it to Voxforge using either your computer or a program like Audacity, or via that old-fashioned instrument, the telephone. There are some other slightly more complicated ways to do it too. I haven’t tried it yet, but it looks like fun.
The other site is called OpenStreetMap.com. The letter writer claimed that in some places it’s more accurate than GoogleMaps, and that it’s a way of updating places that don’t get much attention on Google. I can’t say I’ve found GoogleMaps to be anything but satisfactory, but then maybe I’ve stuck to well-known places, and haven’t tried mapping up in the Andes, or along the Amazon as yet.
OSM seemed a bit complicated to get going on, and, when I found my home address, I noted that the avenue beside my house was listed as a street, so obviously there’s ongoing work to do. Plus it was rather slow to load the Edit module.
A letter to the editor mentioned a couple of sites that sounded interesting, so I’ve just been checking them out. The first is Voxforge.org (not to be confused with Voxforge.com which appears to have been started in 2004 and abandoned). Voxforge’s aim is be a place where you can become part of a service where you offer your voice for free and open source recognition engines.
Voxforge then makes all submitted audio files available under the GPL license, and compiles them into acoustic models for use with Open Source speech recognition engines such as Sphinx, ISIP, Julius and HTK.
GPL, for those, like me, who have never heard of it, is a General Public License. With normal copyright an author has the right to control copies and changes to a work. With a GPL license (also known as "copyleft") a user receives the right to copy and change a work. It’s all part of the process of making the Internet ‘free’, an important word in cyberspace.
Using Voxforge appears to be pretty straightforward: you read and record some text, and upload it to Voxforge using either your computer or a program like Audacity, or via that old-fashioned instrument, the telephone. There are some other slightly more complicated ways to do it too. I haven’t tried it yet, but it looks like fun.
The other site is called OpenStreetMap.com. The letter writer claimed that in some places it’s more accurate than GoogleMaps, and that it’s a way of updating places that don’t get much attention on Google. I can’t say I’ve found GoogleMaps to be anything but satisfactory, but then maybe I’ve stuck to well-known places, and haven’t tried mapping up in the Andes, or along the Amazon as yet.
OSM seemed a bit complicated to get going on, and, when I found my home address, I noted that the avenue beside my house was listed as a street, so obviously there’s ongoing work to do. Plus it was rather slow to load the Edit module.
| 62 |
| Vote |
subscribe to this blog













