Speech to text software for PPC?

[sfetaz]

I have the verizon 6700. I want software on it that can take my voice and convert it to text in something like word or word itself but I cannot find anything that I see does that. Does it exist?

[saqer]

i imagine this would take a large amount of cpu power to achieve and is probably not available for PPC. I have heard of text to speech though

[Torx]

it can barely be done on the pc with the greatest software and fastest cpu power..

[rbeaufor]

Dragon 9

Though its not optimal, I have used Dragon 9 to convert voice memo's I've created on ly PPC. I simply drag them over to my desktop and use Dragon to convert the audio to text.


Admittedly, I've only done it a few times, but it seems to be getting better at it. I use it so infrequently though, I wonder how long it will really take to reach a usable level.

Of note, I heard fonix was good speech to text software, but I never tried it.

[gbm85]

Originally Posted by saqer

i imagine this would take a large amount of cpu power to achieve and is probably not available for PPC. I have heard of text to speech though

That is indeed true. I stumbled across an open-source project that is taking the right approach to this. It involves running a server on a regular PC that recieves a compressed audio stream from a WM client, performs the conversion to text, and sends the resulting text back to the client. While this method is less than ideal, particularly for those without a data plan, it is really the only feasible option.

I'm currently trying to stamp out the bugs in it and make it usable for the lay person, but it will take some time. Debugging a 6-project solution is not an easy task, especially when the documentation is all in a foreign language.

[rsuddath]

Originally Posted by gbm85

That is indeed true. I stumbled across an open-source project that is taking the right approach to this. It involves running a server on a regular PC that recieves a compressed audio stream from a WM client, performs the conversion to text, and sends the resulting text back to the client. While this method is less than ideal, particularly for those without a data plan, it is really the only feasible option.

I'm currently trying to stamp out the bugs in it and make it usable for the lay person, but it will take some time. Debugging a 6-project solution is not an easy task, especially when the documentation is all in a foreign language.

This sounds potentially very useful, I would be happy to test.

What software do you recommend for voice recording, the built in recorder/notes program?

Thanks for your work on this project, I hope that you will update this forum as things procede....

[saqer]

Originally Posted by rbeaufor

Dragon 9

Though its not optimal, I have used Dragon 9 to convert voice memo's I've created on ly PPC. I simply drag them over to my desktop and use Dragon to convert the audio to text.


Admittedly, I've only done it a few times, but it seems to be getting better at it. I use it so infrequently though, I wonder how long it will really take to reach a usable level.

Of note, I heard fonix was good speech to text software, but I never tried it.

Do you have to play the voice memos through the computer's speakers or does Dragon 9 have the option of direct conversion?

[weedahoe]

I was going to also agree about Dragon v8 and v9 as they are very good apps for speech to txt on PC and no they do not require much of a PC at all to run. But no they do not make it for PPC. They key I have found to using these softwares is to articulate and pronouce your works/commands well.

But these type of apps do require more than what the 6700 can provide. Even the HTC Advantage only runs by default at 624mhz

[gbm85]

Originally Posted by rsuddath

This sounds potentially very useful, I would be happy to test.

What software do you recommend for voice recording, the built in recorder/notes program?

Thanks for your work on this project, I hope that you will update this forum as things procede....

The recording is done by the client application. Once the speech is recognized, it gets sent back to the client and appears in the text box.

What I'm doing in order to make this more useful is creating my own client program that can be activated by a hardware button. My app will automatically start recording. Once you press the stop button, the exchange with the server will take place, the interpreted text will automatically be copied to the clipboard, and the client will exit. This will still be a little cludgy, but it is better than the current option.

I would really appreciate any assistance from other C# programmers, especially those who have experience with the SpeechRecognition engine. If you are willing to help, please PM me.

The project I am building upon is licensed under the Microsoft Community License and the LGPL. My client will be licensed under the GPL.