Using TTS when confirming entered number

[Guest Eric Schuyler]

I am evaluating VoiceGuide 5.2.4 for purchase and I am still having problems using TTS text files for the Get Numbers confirmation messages. I have specified text files for all 3 files - before speaking number, after speaking number, and incorrect length.

 

Instead of the text files I specified, I get the "Sound file is the wrong format" error message. All of my other uses of TTS work fine, so I think there is still a problem using TTS in the Get Numbers module.

 

Can someone please confirm whether this is a bug, or help me to understand if I am doing something wrong?

 

Thank you,

Eric Schuyler

eschuyler@emscorp.biz

[SupportTeam]

We confirmed on the test system here that there is problem in v5.2.4 when specifying .txt files in the Get Numbers module's 'Confirm Entered Number' fields.

 

I don't think that using .txt files in the 'Confirm Entered Number' fields was ever supported.

 

We'll look into this and see if we can have this support added in.

 

Do you need to use .txt files in your application or can you perhaps for now just pre-generate these sound files using TTS and specify the .WAV files in those fields instead?

 

PS. this post was split into a new thread as this is a different problem then the one which was discussed in the thread in which you posted originally - http://voiceguide.com/forums/index.php?showtopic=2839

- that thread dealt with verification scripts within the Get Numbers module and that issue was already resolved.

[Guest Eric Schuyler]

Thanks for your fast reply. I will investigate creating WAV files using TTS, but I would definitely prefer being able to use text files. Thanks!

 

I have another question which is not directly related to this topic:

 

I am using the NeoSpeech TTS voices (they are amazing!) with a Dialogic board. The NeoSpeech voices are provided in 8KHz and 16KHz formats. I understand that the Dialogic hardware requires 11KHz files, but I am surprised that the quality degrades so much when downsampling from 16KHz to 11KHz. There is a LOT of background noise and all of the "ss" sounds are particularly bad. I also tried using the 8KHz voice upsampled to 11 KHz, but that isn't any better. I've read the related posts on the forum about this issue, but they all seem to say that the reduced quality can't be improved. I know that the Dialogic board can generate better quality sound, since the prerecorded 11KHz WAV files that come with VoiceGuide sound MUCH better than my TTS voices.

 

Are there any high-quality TTS voices that can be generated directly at 11KHz? I also have the AT&T Natural Voices, which I will experiment with. I don't think they are as good as the NeoSpeech voices, but if the fidelity is better, I might use them.

 

Would installing the "VoiceGuide for Dialogic" patch improve this situation any? In the ReadMe file, there is a statement that bypassing TAPI provides better quality - would this make a difference in the TTS fidelity?

 

Thanks for your help!

 

Regards,

Eric Schuyler

[SupportTeam]

The Dialogic version uses .WAV sound files in 8kHz format - so there will be no downsampling when playing the files. (if 8kHz version of TTS is used).

[Guest Eric Schuyler]

Thanks! I will try the Dialogic version next week.