How To Select .vox Files For Play Module?

[Guest EricGuest]

In http://www.voiceguide.com/vghelp/html/SoundFiles.htm it says,

 

"VoiceGuide for Dialogic can use either WAV files or VOX files. Wave files should be in the following format PCM 8000Hz, 8bit, Mono. VOX files should be in format 4bit ADPCM, 8000Hz."

 

I have VoiceGuide for Dialogic (patched). I have the ability to convert to and produce .vox files with CoolEdit Pro 2.0, and they sound much much clearer and higher quality than files converted to PCM 8000Hz 8Bit Mono files.

 

So I'm trying to figure out how to select these .vox files I've made so that they will be the sound file in the Play Module. The dropdown box will allow me to only select .wav files or text-to-speech .txt files.

 

Thanks.

 

Eric

[ktruk]

Just specify VOX as the filename extention. If you need to 'select' VOX, use the browse button to select WAV, then retype extention from WAV to VOX manually. Its a pain, but it works. Note: you won't be able to play/record VOX via the script-module buttons, they only play WAV. (and only a single WAV in the field too!)

[fluffybunny]

Hi Eric,

 

There should be no need to convert from WAV to VOX, as Voiceguide actually converts the wav file to a vox file on the fly, and the Dialogic plays the vox file not the wav file.

 

Having said this, I have experienced a problem with VG converting WAV's to VOX's on the fly. What happens is that the first time a wav is played VG converts it to a vox and plays the vox. Everytime after this wav is played in the future, VG looks to see if the vox exists and if it does it plays it. This all works fine.

 

However, I installed a system where not all of the WAV files had been played, and so therefore the VOX's, at this point, didn't exist. I assumed they would be created quite happily in the live environment, but from time to time VOX's were created which were 0kb in size. I think it might be something to do with more than one process trying to create the VOX at the same time. i.e. if multiple callers ring in at the same time it screws up the conversion from WAV to VOX, and leaves a 0kb vox file there. Of course VG sees this vox file for all subsequent plays and doesent convert it again, it just plays a 0kb vox file. (i.e. silence!)

 

So, either you can create the vox's yourself and leave them alongside the wav's or you can just make sure every wav file is played at least once by your testing process before you install the system.

 

As far as the script is concerned, if you tell VG to play a WAV file, it will actually play the corresponding VOX file, either by finding the vox if it is already there, or creating it if it isn't. In other words the Dialogic will only ever play the VOX file, not the WAV file, even though the script might ask for the WAV to be played.

 

 

I hope this helps.

 

Cheers

 

Simon

[ktruk]

Just a note on the above: Please be aware that this doubles the disk-space required to manage the files, as every wav file will have a duplicate vox. This is okay if you have a reasonable number of pre-recorded files (or even a large number) but for messaging applications, this can become a significant overhead.

[fluffybunny]

I quite agree, and for messaging applications I would urge you to record a VOX file straight off, rather than a WAV file. (VG works out which format to use for recording simply by looking at the file extension). This way, when listeneing to messages you dont come across the problem with WAV to VOX conversion I mentioned above, as it simply plays the VOX straight off.

[Guest EricGuest]

Thanks folks for your input.

I've discovered that the .vox files that I create directly with CoolEdit Pro 2.0 are playing in slow mode when played through VoiceGuide.

It looks like I still have to record in wav formt, 8 bit 8000 Hz Mono mode, still, and then convert to .vox .

The 8/8000 sound files all have this light hissing in the background.

 

How have you found a way around this?

Thanks.

Eric

[ktruk]

Eric: Light hissing is usually the result of the downsizing/downsampling process and is most noticeable when downsampling from a non-multiple of the frequency, eg: 11.025KHz down to 8KHz, as 11.025 is not a multiple of 8. If the source files are say 32KHz, then downsampled to 8KHz the hiss would be much less, but not eliminated. Also, much better to downsample from 16bit resolution.

 

And, I think Cool Edit deliberately introduces a hiss on sample conversions, as it hides some other effect which is more noticeable if they didn't introduce some hiss. I think this can be turned off in the preferences. Sometimes this effect is compounded by several operations on the one sound-clip, so try and only perform one operation if possible, eg: if you amplify 95% 3 times, you get 3 lots of hiss, but amplify once by 80% only one lot of hiss.

 

You can filter out repetitive hiss using the noise-reduction features quite easily. You may be surprised how good this is! Sometimes its too good and the prompts sound too clean.

 

Finally, don't be mis-led by the sounds coming from your speakers as compared to the phone-line. Always audition your work over the phone to get the real sound, as hiss and conversion artifacts are much less noticeable and you may find it more acceptable.

[fluffybunny]

Incidentally Eric, do you know where I can buy Cool Edit now? I have had a licenced copy for years but had to format my disk and because it was a downloaded product rather than one in a box I didnt have my licence key, so I'm going to have to buy another, as without the licence I have a reduced set of functions.

 

Any idea where I can purchase it from?

 

Cheers

 

Simon

[Guest EricGuest]

I have had mine for a couple of years now. I wasn't aware that it's not being sold direct anymore. A little bit of creative searching on the web might turn up something, perhaps an older licensed version being resold? How about other sound editor programs, like the one that comes with Ahead Nero Suite?