Server side speech to text

Question

Hi,

I'm trying to install a speech recognition engine server side. (non commercial preferred since it's just for experimentation)

The idea is to allow a user to say something from a website then whatever he/she says will show up on the screen (as text)

I've read about many available softwares ranging from Microsoft Speech, Sphinx, Julius etc just not sure which one will perform best and easiest to install. Also do typically do I need to have root permission on my hosting to do this kind of stuff? I'm using a regular shared hosting right now.

Thank you,
Tee

Answer 1

I doubt you will be able to do anything with a shared hosting environment. Speech recognition can be fairly CPU intensive, and a shared hosting environment is 'shared' obviously, and most hosts do not want you stealing all the resources.

Answer 2

I already answered a similar question a hour ago or so. You can use any ASR tool. The two open source ones, htk/julius and sphinx will both work. I used sphinx4 to do this. You can use a shared hosting plan if its just for your own usage but for many users you will need a dedicated server.

You will need root to install lots of the necessary packages. There are also multiple mediums you can choose to receive the audio such as flash/silverlight/google gears/ etc..

There are demos I made at www.speechapi.com that do some simple tasks. Two self promotion posts in a hour. Sorry!