>all the acoutics are supposed to be recorded in 48000Hz.  However all

>the downloads are in 16000 or 8000 Hz.

We downsample all the audio from the VoxForge Speech Submission app to 16kHz and 8kHz.

>I'm pretty sure that you can't mix the two when you are creating your phones list. 

You can't mix audio with different sampling rates when training 'acoustic models'

>julian is expects to read the acoustics coming in at a set range.

You can set the audio sampling rate that Julius expects in the newer version of Juliu

>Also do I need to modify something in the steps 1-10 in order to utilize 16000Hz instead of 48000Hz?

For training acoustic models, no.  All the audio used for training must be the same sampling rate. 

For recognition, you may need to set the sampling rate in Julius. 

The speech audio that you want to recognize must be have the same sampling rate as the audio used to train the acoustic models.

--- (Edited on 12/22/2010 2:27 pm [GMT-0500] by kmaclean) ---

Does this mean that the training of acoustic files can only be at 48000hz?  Or can I train it at 16000 hz?

Since I have only found 16/8K on the site, I assume that I can train in 16Khz.

Also if I train in 48Khz can I utilize a 16khz test acoustics?

Also if I train in 16Khz I assume that I can only test with a lower or equal rating.  Is this correct?

Mike

PS - Happy Holiday

--- (Edited on 12/28/2010 12:47 pm [GMT-0600] by gothrog) ---

--- (Edited on 12/28/2010 12:49 pm [GMT-0600] by gothrog) ---

--- (Edited on 12/28/2010 12:53 pm [GMT-0600] by gothrog) ---

>Does this mean that the training of acoustic files can only be at 48000hz? 

No

>Or can I train it at 16000 hz?

yes, or 8kHz,...

>Since I have only found 16/8K on the site, I assume that I can train in 16Khz.

Yes

>Also if I train in 48Khz can I utilize a 16khz test acoustics?

No - the audio sample rate you train with must be the same sample rate as the audio you want to recognize

>Also if I train in 16Khz I assume that I can only test with a lower or equal rating.

only test/use with audio equal sample rate equal to sample rate you trained with

--- (Edited on 12/30/2010 2:37 pm [GMT-0500] by kmaclean) ---