openrcforums

I was a bit bored today so I made an editor/generator for the voice files in the Taranis.

It is a 0-installation file, just drop it anywhere and run it. It is able to produce all the standard strings in the directory format that is used on Taranis. It is also possible to change any of the spoken sentences and add your own custom sentences.
It is possible to adjust the rate of speech.

Some limitations:
This is a beta. There may well be some bugs.
The application relies heavily on the .NET 4 synthesizer. So no Linux/Mac for this one.
The application only deals with the English set of voice strings.
The application uses the default SAPI voice. If you have anything else than the default English voice set up, you can expect amusing, but hardly usable output.

https://dl.dropboxusercontent.com/u/176 ... Xspeak.exe

This is not a permanent link. It will be open for a couple of days until I have decided if it is worth while to make something more permanent of the editor or if I just scrap it.

I tried it briefly and it seemed to work fine on my Vista 32 bit sys.

Hy,

is this the same sound generator with 5 or 6 languages?

http://212.59.78.77/taranis-sounds-beta1/index.php

Helle

No, my editor/voice generator uses whatever voice you have installed as default on your computer. In practice this usually means the pretty good default voice "Anna".
I am thinking on adding support for selecting any of the TTS languages that are installed and all default stings for all supported languages.

A new version is up that supports switching between all installed TTS voices.
It also supports switching between two lsnguages, but the strings are the same in both. Adding all the language strings is a lot of mechanical work. That will come eventually.

Skickat från min SM-N9005 via Tapatalk

Just trying this out. Seems to make some pretty nice files. I suppose it is not a function of your program rather something that is native to MS and Anna but the only weird thing that I don't like is sometimes she puts the wrong emphasis on certain words that are strung together. Like landing FLAPS? It sort of sounds like a question from her. I am very grateful that there are more knowledgeable people out there than I who are willing to spend time make these sorts of programs. The speed control is genius.

Yes, you are right. It is the speech machinery that determines emphasis . Some things can be corrected by changing the spelling and punctuation (like G-P-S rather than gps), but Anna is very stubborn.

Aren't they all?

Sent from my GT-I9195 using Tapatalk

This is totaly wonderfull

Thanks a lot for your work

A new version is now up. It has full support for English, Swedish and Italian with all strings.
I have added place holders for the non-spoken sounds (Horn, shutter click). The sounds are kind of produced, even if a synthetic voice makes for a bad horn.
I also fixed a problem with the earlier versions that gave the system sound files the names 1.wav, 2.wav rather than 0001.wav, 0002.wav.
French is up next.

Full German and French support added.

Web site is unavailable?

No, I don´t think so. The file is hosted via DropBox, which is very stable.
I just added support for Czech and tested a download. That worked fine.

Sorry my bad. I was using a tablet.

Version 1.0 is up.
All Taranis voice languages are supported. Instructions about installing TTS voices have been added.

Skickat från min SM-N9005 via Tapatalk

cool

Too bad it's not for mac...
I would like my Tx to talk like Darth Vader.

I doubt you'll find a Darth Vader SAPI voice, so you'll have to make that one on your own anyway

How does one export a file from this program?

Ted

Same question, I just tried it and had to edit a few things on the included French template. Closing the program and reopening it, all changes are lost. We should probably be able to save and load a sentence list.

Haven't been able to get decent French files even after loading and selecting supposedly French voices, but that's another problem. The tone changes so the other voice is used, but it still says things with an unuseable English accent.

Or just a single sentence file.

G550Ted wrote:How does one export a file from this program?

Ted

Just generate the sound files for the language you are interested in.
Two text files with the strings used to generate the sound files are written to the same folder as the strings. The same files are read the next time the program is started and the language is loaded. This way you will be able to pick up where you left off.

Kilrah wrote: Haven't been able to get decent French files even after loading and selecting supposedly French voices, but that's another problem. The tone changes so the other voice is used, but it still says things with an unuseable English accent.

In Windows 7 Microsoft supplied a free high quality voice (Anna). But only in English. They published high quality versions of Anna in more than 20 languages that were supposedly going to be available for download and use if you had a correct Windows 7 language version. It sure does not work any longer for my genuine Win7 pro Swedish version. That sucks.
In Windows 8 Anna is gone and replaced with other voices, which are locked down and can not be used through the Speech API. The reason is to force us to pay for commercial alternatives. These are also locked and DRM ptotected and can generally only be used by specialized software. So even if you buy them you are not allowed to use them. That sucks even more. As a result it is harder to get decent speech out of a Windows computer today than it was with my mac in the mid 80ies. This is just sad.
The only half decent free alternative I have found is eSpeak. But I guess that this is the one you have tried and found unusable. It is pretty good in Swedish but worse in other languages.

Skickat från min SM-N9005 via Tapatalk

I guess I don't understand. I have no need of all of the files in the program. I am also not able to delete or otherwise normally manipulate any of the existing files. All I want to do is to create an occasional file or two to export to my Taranis sound folder. Can I do this with this program?

Ted

If someone has any ideas about how to find and install good SAPI5 voices, this is a good place to do so As far as I understand there have been open versions of Anna for all major languages available at some point in time. Perhaps someone knows how to get hold of them.

I'm afraid you are right. Haven't found something useable.

Are there any plans to make it mac compatible?

dvogonen wrote:The only half decent free alternative I have found is eSpeak. But I guess that this is the one you have tried and found unusable.

I haven't actually tried the eSpeak program, but (based on some info I found in a French forum) installed it, and the voices it installs then show up in your program's dropdown list. But even after selecting the french one it plays with English accent and says numbers in English.

That is very strange. I used the eSpeak installer from eSpeaks home page and installed a number of voices, including french. Selecting the voice called eSpeak-fr results in robotic, but clearly French pronunciation. And the numbers are spoken in French.

Skickat från min SM-N9005 via Tapatalk

I think I just found the problem.
I had forgotten to select the synthetic voice before writing files to disk. This meant that the default voice (English) or whatever other voice that had been used previously was used.

I have also added volume controls for the synthetic voice.

The support for synthetic speech for other languages than English has turned out to be almost non-existent.
To still make it possible to generate a sound library I have made a specialized recording application for the sound files that go in the Taranis. It is called OpenTX Recorder. It is nothing else than an automated voice recorder. The advantage is that the program has text prompts for what is to be spoken and then automatically stores the voice files at the right place and in the right name. Judging by my own experience, it is possible to produce voice files at least ten times faster using OpenTX Recorder than with e.g. Audacity.

Version 0.1 can be downloaded from OpenTX home page (http://www.open-tx.org/). But just as the v0.1 implies, it is a very early version and it is guaranteed to be unstable under pressure.