views:

4524

answers:

3

I'd like to build an app that uses voice recognition. I've seen big companies like Google etc implement this feature, but I'm curious about doing it on a start-up level. Anyone looked into this? Are there any tools out there for us to do this?

+4  A: 

The best approach will probably be to:

  1. Record the voice on the phone
  2. Send the recording to a server that runs the speech recognition software
  3. Then return something to the phone to indicate what it should do
John Meagher
That's a lot of data to send. I might try it on the iPhone itself. After all, PCs could do a fair job of this 10 years ago, so perhaps iPhones should be able to now.
Nosredna
This is actually the technique the Google Search app uses
PiPeep
Google encodes the voice in a special way, they don't just send the raw audio data for exactly the reason Nosredna gave.
Alexander Kellett
+5  A: 

If you start here at wikipedia, you'll get a good list engines (http://en.wikipedia.org/wiki/Speech_recognition#Commercial_software.2Fmiddleware)

As I write this (June 24, 2009) it looks to me that are two viable open source solutions

Both have been used in iphone apps, but the iphone friendly source isn't readily available.

As I edit this (8 July, 2009) I recently learned that Loquendo (http://www.loquendo.com/en/) has voice recognition and speech synthesis (ASR & TTS) for the iphone.

JJ Rohrer
@Rohrer, Will Apple approve this if we add any external engines for the voice recognition
Shibin Moideen
@Shibin - I've never heard of such apps being rejected, and I wouldn't expect them to be, either, but your mileage may vary. You can always search around for users of a particular sdk and make sure their apps are actually being published. This would be particularly easy with the commercial sdks.
JJ Rohrer
A: 

Here is a commercial SDK for doing so: http://www.creaceed.com/ceedvocalsdk/

Looks like a wrapper for Julius.

Alexander Kellett